Super Luminal

you can just do python cli.py optimize and then after the db file is created you can then run python cli.py train. You can run train without the costly optuna...

you will have to give the streaming data into the .trade() function with preferably an API call to a data warehouse. and for each incoming tick the real-time bot will...

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...

You can copy all imports, put it at the top. Most of this guy's code is already abstracted properly, so you can just use the class definitions in a python...

write your custom loss function that calculates the metric you want given (y_pred, y_true). Assuming this is for forecasting market data....

mean over axis 2 on [1 5 1] will be reduced to [1 5] yes

TF 1. 1.14/15 has worked on my machine

well you probably have to normalize your data so that you dont have gradient explosions on certain near "out of bounds" data. And any basic vanilla recurrent NN will work...

If this is a univariate time-series forecasting, then the order in which you feed your data into the NN will matter. So what I would also suggest is to stagger...

you can change the model so that it is only trading around a threshold. which agent/agents are you referencing? Some do have a inventory feature.