huseinzol05 / Stock-Prediction-Models

Gathers machine learning and deep learning models for Stock forecasting including trading bots and simulations

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Training Realtime Agent with 5 Minutes Price Interval Dataset Returns Reward 0?

adamfils opened this issue · comments

commented

I tried Training Realtime Agent with 5 Minutes Price Interval Returns Reward 0
Can you please suggest what parameters i have to twerk to train the Realtime Agent with a 5 Minutes Interval Dataset not just 1 day Interval
Thanks

5-min dataset has a lot more "noisey" than day to day aggregates. To really get this to work on intraday data you would need to add (open, high, low) values to your parameters or add some technical analysis to your columns as (ma, 7-day-ma, ema, BB.....) or make more weights jittering per iteration or do sort of elite selection on your population or make your NN-weights "deep" by adding more layers (this would make training slow as this is mainly running on CPU). Since your dataset is different your approach to model selection will be a little different at the end of day. I would say if you want a out of the box solution then either keep training it until you reach some maxima (probably 1000's of iteration as you have noisy data), else I would try the suggestions above.