Open tomjrtsmith opened 12 months ago
Hmm, it's not happening on my machines, but I also just made a lot of changes. Can you send me the version you are using please (or point me to it)?
974 is the live buffer size, and in earlier versions I was restricting training to that size to emulate live conditions. In the latest version (released a few days ago), I switched to an approach where I train on a long time period and save the model; then I load that on subsequent runs and 'tune' it for each individual pair. There are now also several variants that use different algorithms for predicting. I am seeing good results in dry run with TS_Simple_AB (AdaBoost Gradient).
I need to update the documentation, but if you want to try this you have to do the following:
Cheers,
Phil
On Fri, Oct 20, 2023 at 6:16 PM tom s @.***> wrote:
Hey Phil,
TS_Simple is throwing out this error, seems to be a dataframe index length error. I've added one or two additional indicators but that is all. Wondered if any code fixes sprang to mind, seems to run ok on a couple of machines but not others.
Oct 21 00:31:08 instance-20220819-1716 freqtrade_ft3b[590203]: 2023-10-21 00:31:08,362 - freqtrade.strategy.strategy_wrapper - WARNING - Strategy caused the following exception: Length of values (974) does not match length of index (1033)<bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff6e8c9a50>> Oct 21 00:31:08 instance-20220819-1716 freqtrade_ft3b[590203]: 2023-10-21 00:31:08,362 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair XRP/USDT:USDT: Length of values (974) does not match length of index (1033) Oct 21 00:31:08 instance-20220819-1716 freqtrade.strategy.strategy_wrapper
- WARNING - Strategy caused the following exception: Length of values (974) does not match length of index (1033)<bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff6e8c9a50>>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X5Z46D2QQ2PXUVHKAELYAMO6LAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43ASLTON2WKOZRHE2TKMJYHEZTENQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>
This seems to be the central problem I find with the strategies, I'm not sure how to resolve and I'm not sure if its a freqtrade issue or a strat issue, but here you can see it is entering a trade without a buy signal. It does this quite often and it seems to do it for periods at a time. I'm not sure whether the issue is related to lookahead or what causes it.
I see this too. I think the problem/feature is that the model is constantly retraining itself, so the model at the time that it decides to make a trade in real time is not the same as the model that is later used to show historical buy/sell signals. It is sort of a lookahead problem in that the model used in backtesting has been trained on the historical data, so it recognises that it made a mistake. I can't see a way around this (other than restarting the training from scratch each candle) because of the way freqtrade feeds in the data and does not 'remember' the backtest results, so they have to be recalculated each candle, so they change over time.
I'm planning on modifying training in dry/live run situations so that it only re-trains every N candles, let's see if that works - it won't change the trade decisions, but it should provide more realistic backtesting results.
Cheers,
Phil
On Sat, Oct 28, 2023 at 5:46 PM tom s @.***> wrote:
This seems to be the central problem I find with the strategies, I'm not sure how to resolve and I'm not sure if its a freqtrade issue or a strat issue, but here you can see it is entering a trade without a buy signal. It does this quite often and it seems to do it for periods at a time. I'm not sure whether the issue is related to lookahead or what causes it. [image: Screen Shot 2023-10-28 at 2 29 46 PM] https://user-images.githubusercontent.com/99567285/278843638-2970464a-90d8-4e2a-b748-57e92e907faa.png
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1783955419, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X5Z3ZVLTIAYJAF4BFILYBWRN5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTHE2TKNBRHE . You are receiving this because you commented.Message ID: @.***>
FYI, take a look at TS_Coeff, this issue should be fixed. I'll merge the changes into TS_Simple later this week...
Thanks,
Phil
On Mon, Oct 30, 2023 at 7:57 AM Phil Price @.***> wrote:
I see this too. I think the problem/feature is that the model is constantly retraining itself, so the model at the time that it decides to make a trade in real time is not the same as the model that is later used to show historical buy/sell signals. It is sort of a lookahead problem in that the model used in backtesting has been trained on the historical data, so it recognises that it made a mistake. I can't see a way around this (other than restarting the training from scratch each candle) because of the way freqtrade feeds in the data and does not 'remember' the backtest results, so they have to be recalculated each candle, so they change over time.
I'm planning on modifying training in dry/live run situations so that it only re-trains every N candles, let's see if that works - it won't change the trade decisions, but it should provide more realistic backtesting results.
Cheers,
Phil
On Sat, Oct 28, 2023 at 5:46 PM tom s @.***> wrote:
This seems to be the central problem I find with the strategies, I'm not sure how to resolve and I'm not sure if its a freqtrade issue or a strat issue, but here you can see it is entering a trade without a buy signal. It does this quite often and it seems to do it for periods at a time. I'm not sure whether the issue is related to lookahead or what causes it. [image: Screen Shot 2023-10-28 at 2 29 46 PM] https://user-images.githubusercontent.com/99567285/278843638-2970464a-90d8-4e2a-b748-57e92e907faa.png
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1783955419, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X5Z3ZVLTIAYJAF4BFILYBWRN5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTHE2TKNBRHE . You are receiving this because you commented.Message ID: @.***>
Thanks Phil, I was looking at all the updates I'll give it a whirl and report.
On Thu, Nov 16, 2023 at 11:23 AM Nateemma @.***> wrote:
FYI, take a look at TS_Coeff, this issue should be fixed. I'll merge the changes into TS_Simple later this week...
Thanks,
Phil
On Mon, Oct 30, 2023 at 7:57 AM Phil Price @.***> wrote:
I see this too. I think the problem/feature is that the model is constantly retraining itself, so the model at the time that it decides to make a trade in real time is not the same as the model that is later used to show historical buy/sell signals. It is sort of a lookahead problem in that the model used in backtesting has been trained on the historical data, so it recognises that it made a mistake. I can't see a way around this (other than restarting the training from scratch each candle) because of the way freqtrade feeds in the data and does not 'remember' the backtest results, so they have to be recalculated each candle, so they change over time.
I'm planning on modifying training in dry/live run situations so that it only re-trains every N candles, let's see if that works - it won't change the trade decisions, but it should provide more realistic backtesting results.
Cheers,
Phil
On Sat, Oct 28, 2023 at 5:46 PM tom s @.***> wrote:
This seems to be the central problem I find with the strategies, I'm not sure how to resolve and I'm not sure if its a freqtrade issue or a strat issue, but here you can see it is entering a trade without a buy signal. It does this quite often and it seems to do it for periods at a time. I'm not sure whether the issue is related to lookahead or what causes it. [image: Screen Shot 2023-10-28 at 2 29 46 PM] < https://user-images.githubusercontent.com/99567285/278843638-2970464a-90d8-4e2a-b748-57e92e907faa.png>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1783955419>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X5Z3ZVLTIAYJAF4BFILYBWRN5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTHE2TKNBRHE>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1813356343, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNLPTCBJ3VNO2VV2FTTYEU6E3AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJTGM2TMMZUGM . You are receiving this because you authored the thread.Message ID: @.***>
TS_Coeff is throwing this which is a new one. Deprecation warning but I can't find is_sparse.
WARNING - Strategy caused the following exception: could not broadcast input array from shape (1000,) into shape (999,)<bound method IStrategy._analyze_ticker_internal of <TS_Coeff_DWT.TS_Coeff_DWT object at 0xffff74216d50>>
Nov 19 21:12:28 instance-20221222-1628 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair ATOM/USDT:USDT: could not broadcast input array from shape (1000,) into shape (999,)
Nov 19 21:12:28 instance-20221222-1628 freqtrade_ft8a[3607327]: /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/utils/validation.py:757: FutureWarning: is_sparse is deprecated and will be removed in a future version. Check isinstance(dtype, pd.SparseDtype)
instead.
Hmm, I'm not seeing the broadcast error, though I guess it could be caused by changing pair lists. Are you running the latest version?
The deprecation warning is annoying but appears to be OK
Cheers,
Phil
On Sun, Nov 19, 2023 at 1:20 PM tom s @.***> wrote:
TS_Coeff is throwing this which is a new one. Deprecation warning but I can't find is_sparse.
WARNING - Strategy caused the following exception: could not broadcast input array from shape (1000,) into shape (999,)<bound method IStrategy._analyze_ticker_internal of <TS_Coeff_DWT.TS_Coeff_DWT object at 0xffff74216d50>> Nov 19 21:12:28 instance-20221222-1628 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair ATOM/USDT:USDT: could not broadcast input array from shape (1000,) into shape (999,) Nov 19 21:12:28 instance-20221222-1628 freqtrade_ft8a[3607327]: /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/utils/validation.py:757: FutureWarning: is_sparse is deprecated and will be removed in a future version. Check isinstance(dtype, pd.SparseDtype) instead.
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1817980844, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X5YYY7QOTX6P4ZCJU3DYFJZY3AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJXHE4DAOBUGQ . You are receiving this because you commented.Message ID: @.***>
I got rid of the future warning with this below but I'm still getting the broadcast error. Latest versions of all files, TS_Coeff_DWT ran briefly with an old merged json file with some different settings but when I updated to latest everything the broadcast error started again.
import warnings warnings.simplefilter(action='ignore', category=FutureWarning)
WARNING - Unable to analyze candle (OHLCV) data for pair BLZ/USDT:USDT: could not broadcast input array from shape (1025,) into shape (999,)
OK, I'm looking into it, but I'll need you test it for me as I don't get this error.
In the meantime, if you are using ShuffleFilter in your config, try disabling that and see if that works for now.
Cheers,
Phil
On Sun, Nov 19, 2023 at 2:49 PM tom s @.***> wrote:
I got rid of the future warning with this below but I'm still getting the broadcast error. Latest versions of all files, TS_Coeff_DWT ran briefly with an old merged json file with some different settings but when I updated to latest everything the broadcast error started again.
import warnings warnings.simplefilter(action='ignore', category=FutureWarning)
WARNING - Unable to analyze candle (OHLCV) data for pair BLZ/USDT:USDT: could not broadcast input array from shape (1025,) into shape (999,)
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1818003236, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X53XSQSYSGQXMQBZMJLYFKEHVAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJYGAYDGMRTGY . You are receiving this because you commented.Message ID: @.***>
What I'd done is turned the exit signal on in TS_Coeff.json as I didn't think it was working in TS_Coeff_DWT and that triggered the error again but I restarted the bot and it cleared, seems very sensitive to changes
I don't think it's related to that. What I think is happening is that the pairlist must be changing so you get varying sized dataframes supplied and processed in add_latest_prediction(). I maintain a history of the model predictions (which I cannot recalculate because the model is adapting) and restore them to the dataframe. When the pairlist changes (I'm guessing a pair leaves the list and later returns) there could be a mismatch between the sizes of the dataframe and the history. I was only taking care of the case where the dataframe is larger than the history, but I guess it could happen the other way as well.
I've been testing a potential fix overnight with no problems, but I wasn't seeing the issue you had anyway. I'll check it in soon, and would appreciate it if you could verify whether it works for you, since I cannot reproduce the original issue.
Cheers,
Phil
On Sun, Nov 19, 2023 at 11:21 PM tom s @.***> wrote:
What I'd done is turned the exit signal on in TS_Coeff.json as I didn't think it was working in TS_Coeff_DWT and that triggered the error again but I restarted the bot and it cleared, seems very sensitive to changes
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1818365006, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X56Z4SBSWUVIIYXNS7LYFMAGDAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJYGM3DKMBQGY . You are receiving this because you commented.Message ID: @.***>
I saw that happening in the output, definitely runs a lot better but still errors when I reload load the strategy through frequi and needs a restart.
I also started up TS_Simple for the first time in a while and it runs but won't pull in data and it's throwing out this error
freqtrade.strategy.interface - WARNING - Outdated history for pair AGIX/USDT:USDT. Last tick is 18 minutes old
On Tue, Nov 21, 2023 at 3:40 AM Nateemma @.***> wrote:
I don't think it's related to that. What I think is happening is that the pairlist must be changing so you get varying sized dataframes supplied and processed in add_latest_prediction(). I maintain a history of the model predictions (which I cannot recalculate because the model is adapting) and restore them to the dataframe. When the pairlist changes (I'm guessing a pair leaves the list and later returns) there could be a mismatch between the sizes of the dataframe and the history. I was only taking care of the case where the dataframe is larger than the history, but I guess it could happen the other way as well.
I've been testing a potential fix overnight with no problems, but I wasn't seeing the issue you had anyway. I'll check it in soon, and would appreciate it if you could verify whether it works for you, since I cannot reproduce the original issue.
Cheers,
Phil
On Sun, Nov 19, 2023 at 11:21 PM tom s @.***> wrote:
What I'd done is turned the exit signal on in TS_Coeff.json as I didn't think it was working in TS_Coeff_DWT and that triggered the error again but I restarted the bot and it cleared, seems very sensitive to changes
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1818365006>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X56Z4SBSWUVIIYXNS7LYFMAGDAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJYGM3DKMBQGY>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1819192110, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNPIEP6LQTMQIFNJJL3YFNTVPAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZGE4TEMJRGA . You are receiving this because you authored the thread.Message ID: @.***>
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X543JMVDF3OLIQSL6WTYFPPS3AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZHE2TENBZGE . You are receiving this because you commented.Message ID: @.***>
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X543JMVDF3OLIQSL6WTYFPPS3AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZHE2TENBZGE>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNOZC4UZXXKAPDO6IFDYFPYGDAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGAZDANBXGY . You are receiving this because you authored the thread.Message ID: @.***>
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNOZC4UZXXKAPDO6IFDYFPYGDAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGAZDANBXGY>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X553OUVR26IPSNEPIOLYFQOXXAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGE3DCOBRGU . You are receiving this because you commented.Message ID: @.***>
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X553OUVR26IPSNEPIOLYFQOXXAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGE3DCOBRGU>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNMRPRPXK7TZMXAARI3YFQX4RAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGI2DAMZQGA . You are receiving this because you authored the thread.Message ID: @.***>
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNMRPRPXK7TZMXAARI3YFQX4RAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGI2DAMZQGA>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X52VZZM7F6T3SVQ6HKTYFRM4TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGQYDGMZTGA . You are receiving this because you commented.Message ID: @.***>
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X52VZZM7F6T3SVQ6HKTYFRM4TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRQGQYDGMZTGA>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNNWPTMR5UWP2AMDWBTYFTRG5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRRGM3DSNZYGU . You are receiving this because you authored the thread.Message ID: @.***>
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNNWPTMR5UWP2AMDWBTYFTRG5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRRGM3DSNZYGU>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X525LHPOVAHIWW5BWDTYFUTT5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRRHAYDGOBRGY . You are receiving this because you commented.Message ID: @.***>
Exactly the same as the first except before you changed the date format there were two errors, now just one
date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Wed, Nov 22, 2023 at 7:19 PM Nateemma @.***> wrote:
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X525LHPOVAHIWW5BWDTYFUTT5AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRRHAYDGOBRGY>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1822178422, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKHBGUZREBLWKVTQY3YFWKPZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGE3TQNBSGI . You are receiving this because you authored the thread.Message ID: @.***>
apologies. Should be fixed now...
Thanks for testing for me!
Cheers,
Phil
On Wed, Nov 22, 2023 at 12:12 AM tom s @.***> wrote:
Exactly the same as the first except before you changed the date format there were two errors, now just one
date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Wed, Nov 22, 2023 at 7:19 PM Nateemma @.***> wrote:
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1822178422>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNKHBGUZREBLWKVTQY3YFWKPZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGE3TQNBSGI>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1822291942, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X55KLC7LF25PHCCA633YFWXXLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGI4TCOJUGI . You are receiving this because you commented.Message ID: @.***>
Thanks for the amazing algo's Phil.
Date error now fixed but when I download or run test_strat its now throwing this error
freqtrade - ERROR - Start date is after stop date for timerange "20250714-20231122"
On Thu, Nov 23, 2023 at 4:48 AM Nateemma @.***> wrote:
apologies. Should be fixed now...
Thanks for testing for me!
Cheers,
Phil
On Wed, Nov 22, 2023 at 12:12 AM tom s @.***> wrote:
Exactly the same as the first except before you changed the date format there were two errors, now just one
date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Wed, Nov 22, 2023 at 7:19 PM Nateemma @.***> wrote:
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1822178422>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1822291942>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X55KLC7LF25PHCCA633YFWXXLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRSGI4TCOJUGI>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1823023270, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNLA3W5CJXMFPWNPVGTYFYNGJAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRTGAZDGMRXGA . You are receiving this because you authored the thread.Message ID: @.***>
looks like another BSD vs Linux thing. I checked in an update that should work for Linux
Cheers,
Phil
On Wed, Nov 22, 2023 at 9:57 AM tom s @.***> wrote:
Thanks for the amazing algo's Phil.
Date error now fixed but when I download or run test_strat its now throwing this error
freqtrade - ERROR - Start date is after stop date for timerange "20250714-20231122"
On Thu, Nov 23, 2023 at 4:48 AM Nateemma @.***> wrote:
apologies. Should be fixed now...
Thanks for testing for me!
Cheers,
Phil
On Wed, Nov 22, 2023 at 12:12 AM tom s @.***> wrote:
Exactly the same as the first except before you changed the date format there were two errors, now just one
date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Wed, Nov 22, 2023 at 7:19 PM Nateemma @.***> wrote:
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1822178422>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1822291942>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1823023270>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNLA3W5CJXMFPWNPVGTYFYNGJAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRTGAZDGMRXGA>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1823233724, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X567JONPA3SFTDG7Y2LYFY4JLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRTGIZTGNZSGQ . You are receiving this because you commented.Message ID: @.***>
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from '/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
On Thu, Nov 23, 2023 at 3:52 PM Nateemma @.***> wrote:
looks like another BSD vs Linux thing. I checked in an update that should work for Linux
Cheers,
Phil
On Wed, Nov 22, 2023 at 9:57 AM tom s @.***> wrote:
Thanks for the amazing algo's Phil.
Date error now fixed but when I download or run test_strat its now throwing this error
freqtrade - ERROR - Start date is after stop date for timerange "20250714-20231122"
On Thu, Nov 23, 2023 at 4:48 AM Nateemma @.***> wrote:
apologies. Should be fixed now...
Thanks for testing for me!
Cheers,
Phil
On Wed, Nov 22, 2023 at 12:12 AM tom s @.***> wrote:
Exactly the same as the first except before you changed the date format there were two errors, now just one
date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Wed, Nov 22, 2023 at 7:19 PM Nateemma @.***> wrote:
can you show me the error please?
Thanks,
Phil
On Tue, Nov 21, 2023 at 2:31 PM tom s @.***> wrote:
Thanks Phil,
That's cut the errors down from 2 to 1, same error, wonder if it's the date references at the end?
On Wed, Nov 22, 2023 at 6:37 AM Nateemma @.***> wrote:
that's a BSD vs Linux thing.
I updated the scripts so that they should work on Linux as well (at least for that issue)
Cheers,
Phil
On Mon, Nov 20, 2023 at 11:54 PM tom s @.***> wrote:
test_strat has come up with a missing date argument, not sure what it is after
date: invalid option -- 'j' Try 'date --help' for more information. date: invalid option -- 'j' Try 'date --help' for more information. ERR: Missing arguments
On Tue, Nov 21, 2023 at 5:55 PM Nateemma @.***> wrote:
oops, that's a bug. It's looking for user_data/strategies/binanceus/config_binanceus.json (which was there for me)
It really should be user_data/strategies/config/config.json
I'm checking in a fix in now...
Thanks,
Phil
On Mon, Nov 20, 2023 at 7:37 PM tom s @.***> wrote:
Thanks Phil.
Do the permissions need changing for the config_binanceus.json files? The download.sh script can't seem to see them.
On Tue, Nov 21, 2023 at 1:25 PM Nateemma @.***> wrote:
If you delete the .sav file then it will try to train a new model, and will not run backtest (so no buys or sells). You should really run the strat over a long time period (after deleting the .sav file), and then run again (or dry run). Just starting a dry run isn't really enough data to train the model well. I really should update the documentation
Did you change indicators by any chance? If you do that, you will need to regenerate the .sav file
For reference (in case anyone else reads this), the basic process is:
rm -r user_data/strategies/TSPredict/models/TS_Coeff_DWT zsh user_data/strategies/scripts/download.sh -n 600 binanceus # replace exchange name zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Coeff_DWT # this trains a model over 600 days zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Coeff_DWT
backtest strat over last 30 days
You should then be able to do a dry run.
Cheers,
Phil
On Mon, Nov 20, 2023 at 3:11 PM tom s @.***> wrote:
Fixed the error by deleting the .sav file but still no buy signals which seems odd, is this due to training mode?
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1819952491>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820020476>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820161815>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820240300>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1820403330>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821369785>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1821803816>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1822178422>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1822291942>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1823023270>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1823233724>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X567JONPA3SFTDG7Y2LYFY4JLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRTGIZTGNZSGQ>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1823762529, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNLFJ6YOFQEM6UIQ6HTYF225FAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRTG43DENJSHE . You are receiving this because you authored the thread.Message ID: @.***>
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from '/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
Also, as a workaround, you can just specify the dates, e.g.
zsh user_data/strategies/scripts/test_strat.sh -t 20231024- TSPredict TS_Coeff_DWT
(you might need to put 20231024- in quotes, not sure if the Linux shell will do something with the "-")
Thanks,
Phil
On Thu, Nov 23, 2023 at 7:23 AM Phil Price @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from '/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from '/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNPJ4BWTXUUUSLRYQ43YF5TALAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUGYYDSNJYHE . You are receiving this because you authored the thread.Message ID: @.***>
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNPJ4BWTXUUUSLRYQ43YF5TALAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUGYYDSNJYHE>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X565YKPHJIRTK5LYC23YF6J5ZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAYTENJQGY . You are receiving this because you commented.Message ID: @.***>
Yep that fixed it, now backtest throws an attribute error from XGBoost
loading from:
/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/models/TS_Simple/TS_Simple.sav
[19:05:38] WARNING: /workspace/src/learner.cc:553:
If you are loading a serialized model (like pickle in Python, RDS in R)
generated by
older XGBoost, please export the model by calling Booster.save_model
from that version
first, then load it back in current version. See:
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
for more details about differences between saving model and serializing.
backtesting SOL/USDT
2023-11-23 19:05:52,310 - freqtrade - ERROR - Fatal exception! Traceback (most recent call last): File "/home/ubuntu/freqtrade/freqtrade/main.py", line 42, in main return_code = args'func' ^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting backtesting.start() File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1377, in start min_date, max_date = self.backtest_one_strategy(strat, data, timerange) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1294, in backtest_one_strategy preprocessed = self.strategy.advise_all_indicators(data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in advise_all_indicators return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335,
in
On Fri, Nov 24, 2023 at 7:55 AM Nateemma @.***> wrote:
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'... 2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X565YKPHJIRTK5LYC23YF6J5ZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAYTENJQGY>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824823762, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKYV64POLC7E6HXFPLYF6L2TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAZDGNZWGI . You are receiving this because you authored the thread.Message ID: @.***>
Ignore that it just wanted me to do the download, all running now
On Fri, Nov 24, 2023 at 8:10 AM Tom Smith @.***> wrote:
Yep that fixed it, now backtest throws an attribute error from XGBoost
loading from: /home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/models/TS_Simple/TS_Simple.sav [19:05:38] WARNING: /workspace/src/learner.cc:553: If you are loading a serialized model (like pickle in Python, RDS in R) generated by older XGBoost, please export the model by calling
Booster.save_model
from that version first, then load it back in current version. See:https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
for more details about differences between saving model and serializing.
backtesting SOL/USDT
2023-11-23 19:05:52,310 - freqtrade - ERROR - Fatal exception! Traceback (most recent call last): File "/home/ubuntu/freqtrade/freqtrade/main.py", line 42, in main return_code = args'func' ^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting backtesting.start() File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1377, in start min_date, max_date = self.backtest_one_strategy(strat, data, timerange) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1294, in backtest_one_strategy preprocessed = self.strategy.advise_all_indicators(data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in advise_all_indicators return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in
return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy() ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators return self.populate_indicators(dataframe, metadata) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 381, in populate_indicators dataframe = self.add_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 886, in add_predictions dataframe = self.add_jumping_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 770, in add_jumping_predictions self.train_model(base_model, train_data, train_results, False) File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 558, in train_model model.fit(x, results, xgb_model=self.model) File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/core.py", line 620, in inner_f return func(**kwargs) ^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 1006, in fit params = self.get_xgb_params() ^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 708, in get_xgb_params params: Dict[str, Any] = self.get_params() ^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 695, in get_params params.update(cp.class.get_params(cp, deep)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 692, in get_params params = super().get_params(deep) ^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/sklearn/base.py", line 211, in get_params value = getattr(self, key) ^^^^^^^^^^^^^^^^^^ AttributeError: 'XGBModel' object has no attribute 'gpu_id' On Fri, Nov 24, 2023 at 7:55 AM Nateemma @.***> wrote:
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'...
2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X565YKPHJIRTK5LYC23YF6J5ZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAYTENJQGY>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824823762, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKYV64POLC7E6HXFPLYF6L2TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAZDGNZWGI . You are receiving this because you authored the thread.Message ID: @.***>
It backtested ok and starts up but still throws an XGBoost exception it needs the model to be saved a different way
AttributeError: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,787 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,900 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff858a9090>>
On Fri, Nov 24, 2023 at 9:01 AM Tom Smith @.***> wrote:
Ignore that it just wanted me to do the download, all running now
On Fri, Nov 24, 2023 at 8:10 AM Tom Smith @.***> wrote:
Yep that fixed it, now backtest throws an attribute error from XGBoost
loading from: /home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/models/TS_Simple/TS_Simple.sav [19:05:38] WARNING: /workspace/src/learner.cc:553: If you are loading a serialized model (like pickle in Python, RDS in R) generated by older XGBoost, please export the model by calling
Booster.save_model
from that version first, then load it back in current version. See:https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
for more details about differences between saving model and serializing.
backtesting SOL/USDT
2023-11-23 19:05:52,310 - freqtrade - ERROR - Fatal exception! Traceback (most recent call last): File "/home/ubuntu/freqtrade/freqtrade/main.py", line 42, in main return_code = args'func' ^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting backtesting.start() File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1377, in start min_date, max_date = self.backtest_one_strategy(strat, data, timerange)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1294, in backtest_one_strategy preprocessed = self.strategy.advise_all_indicators(data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in advise_all_indicators return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in
return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy() ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators return self.populate_indicators(dataframe, metadata) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 381, in populate_indicators dataframe = self.add_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 886, in add_predictions dataframe = self.add_jumping_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 770, in add_jumping_predictions self.train_model(base_model, train_data, train_results, False) File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 558, in train_model model.fit(x, results, xgb_model=self.model) File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/core.py", line 620, in inner_f return func(**kwargs) ^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 1006, in fit params = self.get_xgb_params() ^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 708, in get_xgb_params params: Dict[str, Any] = self.get_params() ^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 695, in get_params params.update(cp.class.get_params(cp, deep)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 692, in get_params params = super().get_params(deep) ^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/sklearn/base.py", line 211, in get_params value = getattr(self, key) ^^^^^^^^^^^^^^^^^^ AttributeError: 'XGBModel' object has no attribute 'gpu_id' On Fri, Nov 24, 2023 at 7:55 AM Nateemma @.***> wrote:
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'...
2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X565YKPHJIRTK5LYC23YF6J5ZAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAYTENJQGY>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824823762, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKYV64POLC7E6HXFPLYF6L2TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAZDGNZWGI . You are receiving this because you authored the thread.Message ID: @.***>
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
On Thu, Nov 23, 2023 at 2:56 PM tom s @.***> wrote:
It backtested ok and starts up but still throws an XGBoost exception it needs the model to be saved a different way
AttributeError: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,787 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,900 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff858a9090>>
On Fri, Nov 24, 2023 at 9:01 AM Tom Smith @.***> wrote:
Ignore that it just wanted me to do the download, all running now
On Fri, Nov 24, 2023 at 8:10 AM Tom Smith @.***> wrote:
Yep that fixed it, now backtest throws an attribute error from XGBoost
loading from:
/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/models/TS_Simple/TS_Simple.sav
[19:05:38] WARNING: /workspace/src/learner.cc:553: If you are loading a serialized model (like pickle in Python, RDS in R) generated by older XGBoost, please export the model by calling
Booster.save_model
from that version first, then load it back in current version. See:https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
for more details about differences between saving model and serializing.
backtesting SOL/USDT 2023-11-23 19:05:52,310 - freqtrade - ERROR - Fatal exception! Traceback (most recent call last): File "/home/ubuntu/freqtrade/freqtrade/main.py", line 42, in main return_code = args'func' ^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting backtesting.start() File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1377, in start min_date, max_date = self.backtest_one_strategy(strat, data, timerange)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1294, in backtest_one_strategy preprocessed = self.strategy.advise_all_indicators(data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in advise_all_indicators return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in
return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy() ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators return self.populate_indicators(dataframe, metadata) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 381, in populate_indicators dataframe = self.add_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 886, in add_predictions dataframe = self.add_jumping_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 770, in add_jumping_predictions self.train_model(base_model, train_data, train_results, False) File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 558, in train_model model.fit(x, results, xgb_model=self.model) File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/core.py", line 620, in inner_f return func(**kwargs) ^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 1006, in fit params = self.get_xgb_params() ^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 708, in get_xgb_params params: Dict[str, Any] = self.get_params() ^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 695, in get_params params.update(cp.class.get_params(cp, deep)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 692, in get_params params = super().get_params(deep) ^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/sklearn/base.py", line 211, in get_params value = getattr(self, key) ^^^^^^^^^^^^^^^^^^ AttributeError: 'XGBModel' object has no attribute 'gpu_id'
On Fri, Nov 24, 2023 at 7:55 AM Nateemma @.***> wrote:
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO - Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'...
2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824823762>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNKYV64POLC7E6HXFPLYF6L2TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHAZDGNZWGI>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1824980465, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X563PVLFIZHZJHHWRRDYF7IBLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHE4DANBWGU . You are receiving this because you commented.Message ID: @.***>
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
On Thu, Nov 23, 2023 at 2:56 PM tom s @.***> wrote:
It backtested ok and starts up but still throws an XGBoost exception it needs the model to be saved a different way
AttributeError: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,787 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair STORJ/USDT:USDT: 'XGBModel' object has no attribute 'device' Nov 23 22:42:23 instance-20221222-1614 freqtrade_ft7b[1041005]: 2023-11-23 22:42:23,900 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff858a9090>>
On Fri, Nov 24, 2023 at 9:01 AM Tom Smith @.***> wrote:
Ignore that it just wanted me to do the download, all running now
On Fri, Nov 24, 2023 at 8:10 AM Tom Smith @.***> wrote:
Yep that fixed it, now backtest throws an attribute error from XGBoost
loading from:
/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/models/TS_Simple/TS_Simple.sav
[19:05:38] WARNING: /workspace/src/learner.cc:553: If you are loading a serialized model (like pickle in Python, RDS in R) generated by older XGBoost, please export the model by calling
Booster.save_model
from that version first, then load it back in current version. See:https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
for more details about differences between saving model and serializing.
backtesting SOL/USDT 2023-11-23 19:05:52,310 - freqtrade - ERROR - Fatal exception! Traceback (most recent call last): File "/home/ubuntu/freqtrade/freqtrade/main.py", line 42, in main return_code = args'func' ^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/commands/optimize_commands.py", line 58, in start_backtesting backtesting.start() File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1377, in start min_date, max_date = self.backtest_one_strategy(strat, data, timerange)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/optimize/backtesting.py", line 1294, in backtest_one_strategy preprocessed = self.strategy.advise_all_indicators(data) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in advise_all_indicators return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1335, in
return {pair: self.advise_indicators(pair_data.copy(), {'pair': pair}).copy() ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators return self.populate_indicators(dataframe, metadata) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 381, in populate_indicators dataframe = self.add_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 886, in add_predictions dataframe = self.add_jumping_predictions(dataframe) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 770, in add_jumping_predictions self.train_model(base_model, train_data, train_results, False) File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 558, in train_model model.fit(x, results, xgb_model=self.model) File "/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/core.py", line 620, in inner_f return func(**kwargs) ^^^^^^^^^^^^^^ File
"/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 1006, in fit params = self.get_xgb_params() ^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 708, in get_xgb_params params: Dict[str, Any] = self.get_params() ^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 695, in get_params params.update(cp.class.get_params(cp, deep)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File
"/home/ubuntu/.local/lib/python3.11/site-packages/xgboost/sklearn.py", line 692, in get_params params = super().get_params(deep) ^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/ubuntu/.local/lib/python3.11/site-packages/sklearn/base.py", line 211, in get_params value = getattr(self, key) ^^^^^^^^^^^^^^^^^^ AttributeError: 'XGBModel' object has no attribute 'gpu_id'
On Fri, Nov 24, 2023 at 7:55 AM Nateemma @.***> wrote:
Thanks. Can you try the latest version please?
On Thu, Nov 23, 2023 at 10:39 AM tom s @.***> wrote:
The first returned
20231024
The second returned
20231223
On Fri, Nov 24, 2023 at 4:23 AM Nateemma @.***> wrote:
Hmm. Can you do me a favor and type the following 2 commands into a terminal and tell me the results please?:
echo $(date -d "30 days ago " +"%Y%m%d")
echo $(date -d "-30 days ago " +"%Y%m%d")
Thanks,
Phil
On Wed, Nov 22, 2023 at 9:27 PM tom s @.***> wrote:
Still making the same error.
023-11-23 05:14:35,708 - freqtrade.resolvers.iresolver - INFO
Using resolved pairlist StaticPairList from
'/home/ubuntu/freqtrade/freqtrade/plugins/pairlist/StaticPairList.py'...
2023-11-23 05:14:35,715 - freqtrade - ERROR - Start date is after stop date for timerange "20250715-20231123"
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824609589>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1824812506>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824823762>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1824980465>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X563PVLFIZHZJHHWRRDYF7IBLAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRUHE4DANBWGU>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1825974762, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKU2JT4LGX2LEHVPULYGDP4TAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRVHE3TINZWGI . You are receiving this because you authored the thread.Message ID: @.***>
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py", line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py", line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py", line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py", line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py", line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py", line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py", line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value = getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNMUYH5QXLDX2LQJNHDYGEBDJAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGA3TEOBTG4 . You are receiving this because you authored the thread.Message ID: @.***>
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value = getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNMUYH5QXLDX2LQJNHDYGEBDJAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGA3TEOBTG4>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X56IL4IATXQ22P4EJH3YGEFF7AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGA4DSNBUGI . You are receiving this because you commented.Message ID: @.***>
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9569c) [0xffff5f17569c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (1) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9984c) [0xffff5f17984c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (2) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9e8d0) [0xffff5f17e8d0]
On Sat, Nov 25, 2023 at 10:49 AM Nateemma @.***> wrote:
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value = getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X56IL4IATXQ22P4EJH3YGEFF7AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGA4DSNBUGI>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1826102469, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNI2JCHTJLA5DLP5MD3YGEI7XAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGEYDENBWHE . You are receiving this because you authored the thread.Message ID: @.***>
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9569c) [0xffff5f17569c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (1) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9984c) [0xffff5f17984c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (2) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9e8d0) [0xffff5f17e8d0]
On Sat, Nov 25, 2023 at 10:49 AM Nateemma @.***> wrote:
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value = getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X56IL4IATXQ22P4EJH3YGEFF7AVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGA4DSNBUGI>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1826102469, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNI2JCHTJLA5DLP5MD3YGEI7XAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGEYDENBWHE . You are receiving this because you authored the thread.Message ID: @.***>
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9569c)
[0xffff5f17569c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (1)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9984c)
[0xffff5f17984c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (2)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9e8d0)
[0xffff5f17e8d0]
On Sat, Nov 25, 2023 at 10:49 AM Nateemma @.***> wrote:
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value
getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826102469>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/AXXUNNI2JCHTJLA5DLP5MD3YGEI7XAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRWGEYDENBWHE>
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1828296274, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABD4X56UR6Y3ATTCOTHLE7DYGTEKBAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRYGI4TMMRXGQ . You are receiving this because you commented.Message ID: @.***>
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9569c)
[0xffff5f17569c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (1)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9984c)
[0xffff5f17984c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (2)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9e8d0)
[0xffff5f17e8d0]
On Sat, Nov 25, 2023 at 10:49 AM Nateemma @.***> wrote:
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value
getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1826102469>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1828296274>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X56UR6Y3ATTCOTHLE7DYGTEKBAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRYGI4TMMRXGQ>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1828343122, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKJQDZATAPXFQ364R3YGTH6DAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRYGM2DGMJSGI . You are receiving this because you authored the thread.Message ID: @.***>
I get a few of these so not sure what's best to do with them
error: Your local changes to the following files would be overwritten by merge: TSPredict/models/TS_Simple/TS_Simple.sav Please commit your changes or stash them before you merge. Aborting
On Tue, Nov 28, 2023 at 10:15 PM Tom Smith @.***> wrote:
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9569c)
[0xffff5f17569c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (1)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9984c)
[0xffff5f17984c] Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (2)
/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x9e8d0)
[0xffff5f17e8d0]
On Sat, Nov 25, 2023 at 10:49 AM Nateemma @.***> wrote:
this could be an XGBoost version issue (apparently, this parameter was added in version 1.4). Try upgrading:
pip install --upgrade xgboost
Cheers,
Phil
On Fri, Nov 24, 2023 at 1:17 PM tom s @.***> wrote:
I just went and regenerated the model for TS_Simple and ran test_strat again and it started up with the same error - it's an object has no attribute error in SKLearn I save the callbacks
Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,211 - freqtrade.strategy.strategy_wrapper - ERROR - Unexpected error 'XGBModel' object has no attribute 'device' calling <bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff7e2aef90>> Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: Traceback (most recent call last): Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/strategy_wrapper.py", line 27, in wrapper Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return f(*args, **kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 862, in _analyze_ticker_internal Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.analyze_ticker(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 840, in analyze_ticker Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.advise_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File "/home/ubuntu/freqtrade/freqtrade/strategy/interface.py", line 1367, in advise_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return self.populate_indicators(dataframe, metadata) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 362, in populate_indicators Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 867, in add_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: dataframe = self.add_jumping_predictions(dataframe) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 751, in add_jumping_predictions Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: self.train_model(base_model, train_data, train_results, False) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/user_data/strategies/nateemadev/TSPredict/TS_Simple.py",
line 539, in train_model Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: model.fit(x, results, xgb_model=self.model) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/core.py",
line 729, in inner_f Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: return func(**kwargs) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 1069, in fit Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = self.get_xgb_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 797, in get_xgb_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params: Dict[str, Any] = self.get_params() Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 784, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params.update(cp.class.get_params(cp, deep)) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/sklearn.py",
line 781, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: params = super().get_params(deep) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: File
"/home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/sklearn/base.py",
line 211, in get_params Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: value
getattr(self, key) Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: ^^^^^^^^^^^^^^^^^^ Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: AttributeError: 'XGBModel' object has no attribute 'device' Nov 24 21:10:31 instance-20221222-1614 freqtrade_ft7b[1056653]: 2023-11-24 21:10:31,212 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair SSV/USDT:USDT: 'XGBModel' object has no attribute 'device'
On Sat, Nov 25, 2023 at 9:42 AM Nateemma @.***> wrote:
In that case. just regenerate the models:
rm user_data/strategies/TSPredict/models/TS_Simple/TS_Simple.sav
zsh user_data/strategies/scripts/test_strat.sh -n 600 TSPredict TS_Simple
zsh user_data/strategies/scripts/test_strat.sh -n 30 TSPredict TS_Simple
The 600 and 30 are just for illustration, just use a long time period to generate the model.
As an aside, if you want to regenerate all of the models, do this:
rm -r user_data/strategies/TSPredict/models/TS_Simple zsh user_data/strategies/scripts/test_group.sh -n 600 TSPredict "TS_Simple"
Cheers,
Phil
On Fri, Nov 24, 2023 at 12:05 PM tom s @.***> wrote:
Won't that recreate the problem? I've added some indicators to TS_Simple so they won't work with the regenerated model will they?
On Sat, Nov 25, 2023 at 7:15 AM Nateemma @.***> wrote:
Oh, that's just an old model - I don't usually run the base classes (TS_Simple & TS_Coef). I re-generated and checked them in
Cheers,
Phil
Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826072837>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826089442>,
or unsubscribe <
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub <
https://github.com/nateemma/strategies/issues/38#issuecomment-1826102469>,
or unsubscribe <
. You are receiving this because you authored the thread.Message ID: @.***>
— Reply to this email directly, view it on GitHub < https://github.com/nateemma/strategies/issues/38#issuecomment-1828296274>,
or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABD4X56UR6Y3ATTCOTHLE7DYGTEKBAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRYGI4TMMRXGQ>
. You are receiving this because you commented.Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1828343122, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNKJQDZATAPXFQ364R3YGTH6DAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMRYGM2DGMJSGI . You are receiving this because you authored the thread.Message ID: @.***>
yes, it's OK to overwrite them (as long as you've trained over a decent time period). The saved models are really just a kick start mechanism - there isn't enough data in the live stream to train the regression algorithm in real time, so I use those models as a starting point and then train incrementally on the real time data. If you change indicators (or XGBoost changes model format) then you have to retrain.
FYI, TS_Wavelet (and TS_Wavelet_DWTA) can be tried now - no saved models :-)
Cheers,
Phil
On Tue, Nov 28, 2023 at 1:16 AM tom s @.***> wrote:
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
Message ID: @.***>
TS_Simple is throwing out this error, I removed all indicators looks like an XGBoost setting
Nov 30 22:17:36 instance-20220819-1716 freqtrade_ft3c[1149165]: [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730] Nov 30 22:17:36 instance-20220819-1716 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair FITFI/USDT:USDT: [22:17:36] /workspace/src/learner.cc:1449: Check failed: learner_modelparam.num_feature == p_fmat->Info().numcol (31 vs. 33) : Number of columns does not match number of features in booster.#012Stack trace:#012 [bt] (0) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x354994) [0xffff7db04994]#012 [bt] (1) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x371b38) [0xffff7db21b38]#012 [bt] (2) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x3890c8) [0xffff7db390c8]#012 [bt] (3) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(XGBoosterUpdateOneIter+0x54) [0xffff7d825d9c]#012 [bt] (4) /lib/aarch64-linux-gnu/libffi.so.8(+0x6e10) [0xffffa7926e10]#012 [bt] (5) /lib/aarch64-linux-gnu/libffi.so.8(+0x3a94) [0xffffa7923a94]#012 [bt] (6) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x12994) [0xffffa7952994]#012 [bt] (7) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x10cac) [0xffffa7950cac]#012 [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730]#012
On Wed, Nov 29, 2023 at 6:11 AM Nateemma @.***> wrote:
yes, it's OK to overwrite them (as long as you've trained over a decent time period). The saved models are really just a kick start mechanism - there isn't enough data in the live stream to train the regression algorithm in real time, so I use those models as a starting point and then train incrementally on the real time data. If you change indicators (or XGBoost changes model format) then you have to retrain.
FYI, TS_Wavelet (and TS_Wavelet_DWTA) can be tried now - no saved models :-)
Cheers,
Phil
On Tue, Nov 28, 2023 at 1:16 AM tom s @.***> wrote:
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1830320592, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNI7SDIODQJENUPJ35TYGYLLFAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZQGMZDANJZGI . You are receiving this because you authored the thread.Message ID: @.***>
Scratch that sorry just needed training.
On Fri, Dec 1, 2023 at 11:20 AM Tom Smith @.***> wrote:
TS_Simple is throwing out this error, I removed all indicators looks like an XGBoost setting
Nov 30 22:17:36 instance-20220819-1716 freqtrade_ft3c[1149165]: [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730] Nov 30 22:17:36 instance-20220819-1716 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair FITFI/USDT:USDT: [22:17:36] /workspace/src/learner.cc:1449: Check failed: learner_modelparam.num_feature == p_fmat->Info().numcol (31 vs. 33) : Number of columns does not match number of features in booster.#012Stack trace:#012 [bt] (0) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x354994) [0xffff7db04994]#012 [bt] (1) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x371b38) [0xffff7db21b38]#012 [bt] (2) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x3890c8) [0xffff7db390c8]#012 [bt] (3) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(XGBoosterUpdateOneIter+0x54) [0xffff7d825d9c]#012 [bt] (4) /lib/aarch64-linux-gnu/libffi.so.8(+0x6e10) [0xffffa7926e10]#012 [bt] (5) /lib/aarch64-linux-gnu/libffi.so.8(+0x3a94) [0xffffa7923a94]#012 [bt] (6) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x12994) [0xffffa7952994]#012 [bt] (7) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x10cac) [0xffffa7950cac]#012 [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730]#012
On Wed, Nov 29, 2023 at 6:11 AM Nateemma @.***> wrote:
yes, it's OK to overwrite them (as long as you've trained over a decent time period). The saved models are really just a kick start mechanism - there isn't enough data in the live stream to train the regression algorithm in real time, so I use those models as a starting point and then train incrementally on the real time data. If you change indicators (or XGBoost changes model format) then you have to retrain.
FYI, TS_Wavelet (and TS_Wavelet_DWTA) can be tried now - no saved models :-)
Cheers,
Phil
On Tue, Nov 28, 2023 at 1:16 AM tom s @.***> wrote:
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1830320592, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNI7SDIODQJENUPJ35TYGYLLFAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZQGMZDANJZGI . You are receiving this because you authored the thread.Message ID: @.***>
Phil what are your thoughts on the lookahead setting fat 9 or TS_Wavelet? It seems to be making a few more false buys where there's no buy signal than the other strats where the lookahead is set to 6.
I'm also not sure it's obeying all its exit signals and wondered whether the cexit settings might be getting in the way of the prediction model.
You can see here it's done one trade ignored exit signals and then a false buy and I don't know if that was happening when it was set at 6.
On Fri, Dec 1, 2023 at 11:57 AM Tom Smith @.***> wrote:
Scratch that sorry just needed training.
On Fri, Dec 1, 2023 at 11:20 AM Tom Smith @.***> wrote:
TS_Simple is throwing out this error, I removed all indicators looks like an XGBoost setting
Nov 30 22:17:36 instance-20220819-1716 freqtrade_ft3c[1149165]: [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730] Nov 30 22:17:36 instance-20220819-1716 freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair FITFI/USDT:USDT: [22:17:36] /workspace/src/learner.cc:1449: Check failed: learner_modelparam.num_feature == p_fmat->Info().numcol (31 vs. 33) : Number of columns does not match number of features in booster.#012Stack trace:#012 [bt] (0) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x354994) [0xffff7db04994]#012 [bt] (1) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x371b38) [0xffff7db21b38]#012 [bt] (2) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(+0x3890c8) [0xffff7db390c8]#012 [bt] (3) /home/ubuntu/freqtrade/.venv/lib/python3.11/site-packages/xgboost/lib/libxgboost.so(XGBoosterUpdateOneIter+0x54) [0xffff7d825d9c]#012 [bt] (4) /lib/aarch64-linux-gnu/libffi.so.8(+0x6e10) [0xffffa7926e10]#012 [bt] (5) /lib/aarch64-linux-gnu/libffi.so.8(+0x3a94) [0xffffa7923a94]#012 [bt] (6) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x12994) [0xffffa7952994]#012 [bt] (7) /usr/lib/python3.11/lib-dynload/ ctypes.cpython-311-aarch64-linux-gnu.so(+0x10cac) [0xffffa7950cac]#012 [bt] (8) /home/ubuntu/freqtrade/.venv/bin/python3.11(_PyObject_MakeTpCall+0x240) [0x490730]#012
On Wed, Nov 29, 2023 at 6:11 AM Nateemma @.***> wrote:
yes, it's OK to overwrite them (as long as you've trained over a decent time period). The saved models are really just a kick start mechanism - there isn't enough data in the live stream to train the regression algorithm in real time, so I use those models as a starting point and then train incrementally on the real time data. If you change indicators (or XGBoost changes model format) then you have to retrain.
FYI, TS_Wavelet (and TS_Wavelet_DWTA) can be tried now - no saved models :-)
Cheers,
Phil
On Tue, Nov 28, 2023 at 1:16 AM tom s @.***> wrote:
Phil what's the deal when you regenerate models? Is it ok to overwrite existing trained models?
On Tue, Nov 28, 2023 at 6:56 AM Nateemma @.***> wrote:
Looks like you need to regenerate the models for TS_Coeff* (I’m guessing you have extra indicators)
For TC_Wavelet, I’m still developing that; not ready for general use yet
Cheers
Phil
On Mon, Nov 27, 2023 at 9:25 AM tom s @.***> wrote:
Updated TS_Coeff_DWT and it started throwing this error, was running fine previously.
I've also started TC_Wavelet and it runs and generates awesome buy signals but makes no trades, does it need backtest data as well?
2023-11-27 17:18:12,622 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair GRT/USDT:USDT: X has 188 features, but SGDRegressor is expecting 193 features as input.
On Sat, Nov 25, 2023 at 12:21 PM Tom Smith @.***> wrote:
I was running 2.01 (error) upgraded to 2.02 (error) so tried installing 1.50 and got a stack trace error
I wondered if it could be related to the way joblib.dump(self.model, model_path)works
https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Stack trace: Nov 24 23:03:16 instance-20221222-1614 freqtrade_ft7b[1057366]: [bt] (0)
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1830320592, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNI7SDIODQJENUPJ35TYGYLLFAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZQGMZDANJZGI . You are receiving this because you authored the thread.Message ID: @.***>
By coincidence, I just changed the default lookahead to 6.
As an aside, if you want to change lookahead for the Wavelet strategies, do that in TS_Wavelet.py, not TSPredict.py - TSPredict.py will also affect the other strategies, and if you change it, you need to re-generate the saved models again (because the training data is now different).
I have noticed that if you have your order types set to "limit" in your config file then it often times out on both buys and sells. If you just look at the charts, it appears as if the strategy ignored buy/sell signals. The file user_data/strategies/config/config.json is set up for "market" trades if you need an example
Hope that helps,
Cheers,
Phil
On Wed, Dec 6, 2023 at 12:55 AM tom s @.***> wrote:
Phil what are your thoughts on the lookahead setting fat 9 or TS_Wavelet? It seems to be making a few more false buys where there's no buy signal than the other strats where the lookahead is set to 6.
I'm also not sure it's obeying all its exit signals and wondered whether the cexit settings might be getting in the way of the prediction model.
You can see here it's done one trade ignored exit signals and then a false buy and I don't know if that was happening when it was set at 6.
On Fri, Dec
Message ID: @.***>
Phil I'm running TS_Wavelet with leverage but if I put the leverage callback into TS_Predict it won't seem to work with subclasses. Any idea why this is?
On Thu, Dec 7, 2023 at 5:52 AM Nateemma @.***> wrote:
By coincidence, I just changed the default lookahead to 6.
As an aside, if you want to change lookahead for the Wavelet strategies, do that in TS_Wavelet.py, not TSPredict.py - TSPredict.py will also affect the other strategies, and if you change it, you need to re-generate the saved models again (because the training data is now different).
I have noticed that if you have your order types set to "limit" in your config file then it often times out on both buys and sells. If you just look at the charts, it appears as if the strategy ignored buy/sell signals. The file user_data/strategies/config/config.json is set up for "market" trades if you need an example
Hope that helps,
Cheers,
Phil
On Wed, Dec 6, 2023 at 12:55 AM tom s @.***> wrote:
Phil what are your thoughts on the lookahead setting fat 9 or TS_Wavelet? It seems to be making a few more false buys where there's no buy signal than the other strats where the lookahead is set to 6.
I'm also not sure it's obeying all its exit signals and wondered whether the cexit settings might be getting in the way of the prediction model.
You can see here it's done one trade ignored exit signals and then a false buy and I don't know if that was happening when it was set at 6.
On Fri, Dec
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1843283129, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNM3MKJIIHYQPHG7GWLYICPELAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBTGI4DGMJSHE . You are receiving this because you authored the thread.Message ID: @.***>
have you verified that the callback is running?
If it's not, are you using the correct config file? I think you need to set trading_mode and (perhaps) margin_mode
You can also try setting the global leverage parameter (at class level) to something > 1.0 and see if it runs
Cheers,
Phil
On Thu, Dec 7, 2023 at 3:13 PM tom s @.***> wrote:
Phil I'm running TS_Wavelet with leverage but if I put the leverage callback into TS_Predict it won't seem to work with subclasses. Any idea why this is?
Message ID: @.***>
Thanks Phil,
I'm also trying to write a short condition for TS_Wavelet but I'm not sure it will work, I've used the target_loss dataframe in model_cond but I'm not sure that will give it a short entry position. .
conditions = [] dataframe.loc[:, 'enter_tag'] = '' if self.training_mode: dataframe['enter_short'] = 0 return dataframe if self.enable_short_guards.value:
conditions.append(dataframe['fisher_wr'] < self.short_guard_fwr.value)
if 'volume' in dataframe: conditions.append(dataframe['volume'] > 0)
fwr_short_cond = ( (dataframe['fisher_wr'] > 0.98) )
model_cond = ( (
qtpylib.crossed_below(dataframe['predicted_gain'], dataframe['target_loss'])
dataframe['target_profit'].shift()) &
)
dataframe['target_profit'])
)
conditions.append(model_cond)
dataframe.loc[fwr_short_cond, 'enter_tag'] += 'fwr_entry ' dataframe.loc[model_short_cond, 'enter_tag'] += 'model_entry '
if conditions: dataframe.loc[reduce(lambda x, y: x & y, conditions), 'enter_short'] = 1
On Fri, Dec 8, 2023 at 12:47 PM Nateemma @.***> wrote:
have you verified that the callback is running?
If it's not, are you using the correct config file? I think you need to set trading_mode and (perhaps) margin_mode
You can also try setting the global leverage parameter (at class level) to something > 1.0 and see if it runs
Cheers,
Phil
On Thu, Dec 7, 2023 at 3:13 PM tom s @.***> wrote:
Phil I'm running TS_Wavelet with leverage but if I put the leverage callback into TS_Predict it won't seem to work with subclasses. Any idea why this is?
Message ID: @.***>
— Reply to this email directly, view it on GitHub https://github.com/nateemma/strategies/issues/38#issuecomment-1846276640, or unsubscribe https://github.com/notifications/unsubscribe-auth/AXXUNNNCNLVUNAGU3X56LC3YIJIRFAVCNFSM6AAAAAA6JYNPWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBWGI3TMNRUGA . You are receiving this because you authored the thread.Message ID: @.***>
Hey Phil,
TS_Simple is throwing out this error, seems to be a dataframe index length error. I've added one or two additional indicators but that is all. Wondered if any code fixes sprang to mind, seems to run ok on a couple of machines but not others.
Oct 21 00:31:08 instance-20220819-1716 freqtrade_ft3b[590203]: 2023-10-21 00:31:08,362 - freqtrade.strategy.strategy_wrapper - WARNING - Strategy caused the following exception: Length of values (974) does not match length of index (1033)<bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff6e8c9a50>> Oct 21 00:31:08 instance-20220819-1716 freqtrade_ft3b[590203]: 2023-10-21 00:31:08,362 - freqtrade.strategy.interface - WARNING - Unable to analyze candle (OHLCV) data for pair XRP/USDT:USDT: Length of values (974) does not match length of index (1033) Oct 21 00:31:08 instance-20220819-1716 freqtrade.strategy.strategy_wrapper - WARNING - Strategy caused the following exception: Length of values (974) does not match length of index (1033)<bound method IStrategy._analyze_ticker_internal of <TS_Simple.TS_Simple object at 0xffff6e8c9a50>>