How to use other parameters as reward for the RL models to train?

[Mr-Nobody-dey]

In the examples mentioned in the repository, the models use portfolio value as reward.

What if i want to use other parameter as reward?

1. Do I need to redefine/build a new RL model for that task?
2. Or their is some other easy way?

Describe the solution you'd like Please provide a solution in the tutorial repository.

Describe alternatives you've considered NA

Additional context NA

[mmmarchetti]

Hello @Mr-Nobody-dey! If you need to modify the reward calculation parameter, you can create a customized environment by cloning the project and making changes to the finrl/meta folder. Actually, the default environment does not offer this option.

[Mr-Nobody-dey]

Ok. Thank you for your comment.