Leak Feature Doesn't Effect Performance about finrl HOT 3 CLOSED

First, thank you for this awesome project.

I try to add some leaky features to the model to see if it is capable of exploiting that kind of features. Normally I would expect to see some incredible result, because basically I'm telling the future to the model. But; model still works poorly. Why would that be you think? Is it maybe because of parameters of the model?

Also, in some models I see that, in some iterations rewards became 0 with these leaky features. So, some models choose not the operate instead of exploit these future information.

Thanks.

Good idea, what are the leaky features?

from finrl.

@BruceYanghy first of all I changed 'daily_return' to this:
df.close.shift(-10).pct_change(1).rolling(10).mean().fillna(0)
and also created similar features and finally added local minimum and maximum points. All these features can not be known in the present time.

Local minimum and maximum informations are extreme leakeage features. Even if you implement so much simple trading strategy it would make great impact (but it can not be known). I would expect the RL model to exploit these.

from finrl.

Thanks for you suggestions!

from finrl.