Open chatzich opened 6 years ago
Those shown relativeProfit results are really not backtest results.
There are formulas to convert backtest results to a number that is more accurate to evaluate actual strategy success.
You can select which one to use at settings.generations.interpreteBacktestProfit
and you can check them at the beggining of evaluate/gekko/backtest.py
.
I have not ran sufficient GAs to actually say that those interpretations guarantee better evolutions for sure, and that's why you can select which one you want ('v1' returns real backtest result).
cheers!
Locale files contains data used to construct those plots at -w
mode. They resumes evolution statistics over time for each locale.
@Gab0 thanks for the answer but I can't manage to select the v1 interpreter ... which option shall I use?
At Settings.py
, look for interpreteBacktestProfit
key ;)
@Gab0 I made the changes you insist and thank you but I can't still take the actual profit
Relative profit on evolution dataset: 130.711 Current parameters are viable. 130.711 final profit ~~~~ { "interval": 8.4, "thresholds": { "high": 45.0, "low": 30.299999999999997, "persistence": 4.6 } } interval = 8.4 [thresholds] high = 45.0 low = 30.299999999999997 persistence = 4.6
Hi Ironexmaiden,
I am not sure it will help... Did you configure your backtest with a trading fee of 0% ? AFAIK, Japonicus uses 0% fees so if you backtested with Gekko default settings (0.25% fees), less than 80 trades will make your profit drop from 18% to -19%.
Hope it helps! Xavier
@ironexmaiden Hey, size of candles for backtest are aways in minutes. Did you set that option to 'v1'? Should get transparent results... if still has problems send some logs.. @xaviermiss japonicus uses gekko default tradding fees. cheers!
1.The report file is on logs/STRATEGY-MARKET-USDT-BTC-NUMBER/japonicus.log 2.And if it is how can I read the report? There are several locales which report with several strategy parameters but when I backtesting these parameters the result is different than the japonicus.log's one
`Testing Strategy of Locale3 @ EPOCH 350:
Test on random candles... relativeProfit: -13.513 nbTrades: 5.0 sharpe: -4.85 2018-01-27 01:40:29 to 2018-04-27 01:40:29 BTC/USDT @poloniex
Test on random candles... relativeProfit: 58.278 nbTrades: 3.0 sharpe: 0.00 2018-01-03 00:42:59 to 2018-04-03 00:42:59 BTC/USDT @poloniex
Test on random candles... relativeProfit: -18.360 nbTrades: 5.0 sharpe: -0.88 2018-01-16 01:05:00 to 2018-04-16 01:05:00 BTC/USDT @poloniex
Test on random candles... relativeProfit: 31.976 nbTrades: 3.0 sharpe: 0.00 2018-01-19 03:51:25 to 2018-04-19 03:51:25 BTC/USDT @poloniex
Test on random candles... relativeProfit: -6.821 nbTrades: 5.0 sharpe: -1.18 2018-01-21 11:58:20 to 2018-04-21 11:58:20 BTC/USDT @poloniex
Test on random candles... relativeProfit: 65.555 nbTrades: 7.0 sharpe: 1.93 2018-01-01 17:05:13 to 2018-04-01 17:05:13 BTC/USDT @poloniex
Test on random candles... relativeProfit: -11.220 nbTrades: 3.0 sharpe: 0.00 2018-01-28 23:41:24 to 2018-04-28 23:41:24 BTC/USDT @poloniex
Test on random candles... relativeProfit: 31.021 nbTrades: 5.0 sharpe: -1.18 2018-01-20 02:28:03 to 2018-04-20 02:28:03 BTC/USDT @poloniex
Test on random candles... relativeProfit: -6.000 nbTrades: 9.0 sharpe: -1.55 2018-01-02 20:07:46 to 2018-04-02 20:07:46 BTC/USDT @poloniex
Test on random candles... relativeProfit: -9.635 nbTrades: 5.0 sharpe: -1.18 2018-01-08 00:08:56 to 2018-04-08 00:08:56 BTC/USDT @poloniex
Test on random candles... relativeProfit: -3.667 nbTrades: 3.0 sharpe: 0.00 2018-01-22 23:13:17 to 2018-04-22 23:13:17 BTC/USDT @poloniex
Test on random candles... relativeProfit: -1.573 nbTrades: 5.0 sharpe: -1.18 2018-01-20 09:48:25 to 2018-04-20 09:48:25 BTC/USDT @poloniex
Relative profit on evolution dataset: 116.043 Current parameters are viable. 116.043 final profit ~~~~ { "interval": 8.4, "thresholds": { "high": 64.95, "low": 18.0, "persistence": 9.64 } } interval = 8.4 [thresholds] high = 64.95 low = 18.0 persistence = 9.64 `
reports me on back test ` simulated profit:-18.03484%
`
Thanks