Is there any possibility to use my own data?

[JafferWilson]

Hello, as I see you are using the gekko for data and prediction. How I can use your repo for producing result base on my own dataset from another platform and display the output in the command window or in jupyter notebook? Please let me know

[Gab0]

Candles can be fed via the http listener, or just wrote up in candles file, which is expected to by a 3d python array (batch, small number of sequential candles, OHLCV). Thus you have two options:

a) Replicate neuralloader.js gekko strategy funcionality on your desired platform. b) Write the array of candles directly on candles (and maybe on evalcandles).

Any question just ask... cheers!

[sakirun]

Hi, Gab0

I can't find example for input the candle using http post. I've searched for neuralloader.js but can't find it. can you give me example of data format to post using http.

[Gab0]

Hey @sakirun, this project is extremely WIP and I don't reccomend it unless you're a hardcore researcher xD

The gekko strat neuraloader.js and the indicator nncandle.js were really missing, added.

There you can find the http request outline, which is like this:

addr = localhost:5000/candleinput
payload = {'candle': [O, H, L, C, V]}

This request can be done from any platform, of course... cheers!

[sakirun]

Thanks for the quick reply and update,

I think there's typo in: @app.route('/candleinput', methods=['POST']) def storeCandlestick(): Candles = json.loads(request.data) app.candleBank[app.currentBank].attach(Candles['candles']) print(Candles)

should be: @app.route('/candleinput', methods=['POST']) def storeCandlestick(): Candles = json.loads(request.data) app.candleBanks[app.currentBank].attach(Candles['candles']) print(Candles)

[Gab0]

Thanks, I'll update in a while. Feel free to make PRs with this and/or more ideas, if you manage to understand the workings of this... I will tell in advance that I didn't get predictions results reliable enough to apply on my own live trading usage.

[sakirun]

Great,

I try to manage this project to predict binary option candlestick, still working. I will make PR if i can make it work good enough