search

AI4Finance-Foundation / FinRL

FinRL: Financial Reinforcement Learning. 🔥
https://ai4finance.org
MIT License
9.93k stars 2.4k forks source link

Is the Sharp Ratio function optimizable for any data? #458

Closed ghaffari903 closed 2 years ago

ghaffari903 commented 2 years ago

1- Is the Sharp Ratio function optimizable for any data?

2- By what parameter do we increase the number of training episodes (both stable-baselines3 an elegant)?

Burntt commented 2 years ago

  1. Yes, you can optimize for Sharpe ratio. During hyperparameter tuning, you can have the target to be Sharpe ratio.

  2. break_step. This is the variable that determines the total amount of 'steps' in any of the trading environments. For example: if you have a timeseries, hourly with 10.000 points, and you set break_step to 20.000, the full datasets is passed twice.