What type of data can the model handle?
I can see that the data processor you provided ouputs a csv with yahoo finance data in 1d intervals, can the model handle 1 hour and 1 minute intervals too?
Yes — the model can be extended to 1-hour or 1-minute intervals, but it requires significant adaptation. You’ll need to adjust the interval configuration, redefine lookback windows, rescale features, and redesign the reward to account for higher trading frequency and costs. Without these changes, performance will likely degrade due to noise, insufficient context, and over-reactivity. The key reason is that the model was originally trained and tuned on daily data, so intraday use demands careful retuning and retraining. You can check how everything was organised in this repo https://github.com/AdilzhanB/StockRL.