FinRL: The first open-source project for financial reinforcement learning
1–10 of 58 posts
Re: FinRL: The first open-source project for financial reinforcement learning
#2As someone with 10+ years of exp in the industry, I had no clue what this was actually doing after reading the github page. Had to dive through the examples and videos.
The alphas showcased seem weak: 1.5 max gross SR in sample, unresidualized, unconstrained and standalone.
That would need to be blended with a hundred un correlated others to have a decent portfolio.
Re: FinRL: The first open-source project for financial reinforcement learning
#3I might be able to use these algorithms in my automated trading system: https://tradecast.one. I'm not sure how well they perform, but something to consider.
Re: FinRL: The first open-source project for financial reinforcement learning
#4I'm curious how efficient is RL for trading. Given that RL is extremely data-hungry I'd say that this approach is likely to result in an overfit model. Does anybody have positive experiences with this?
Re: FinRL: The first open-source project for financial reinforcement learning
#5This is self defeating if you open source an approach like this and it is effective, then the laws of arbitrage says that the market will remove any additional benefits of using it.
Re: FinRL: The first open-source project for financial reinforcement learning
#6This is self defeating if you open source an approach like this and it is effective, then the laws of arbitrage says that the market will remove any additional benefits of using it.
It would have to be the DOMINANT approach for that to happen.
Re: FinRL: The first open-source project for financial reinforcement learning
#7As someone with 10+ years of exp in the industry, I had no clue what this was actually doing after reading the github page. Had to dive through the examples and videos. The alphas showcased seem weak: 1.5 max gross SR in sample, unresidualized, unconstrained and standalone. That would need to be blended with a hundred un correlated others to have a decent portfolio.
What does unresidualized mean?
What do the need to do to the software to “residualize” it? If that’s even the right expression.
Re: FinRL: The first open-source project for financial reinforcement learning
#8As someone with 10+ years of exp in the industry, I had no clue what this was actually doing after reading the github page. Had to dive through the examples and videos. The alphas showcased seem weak: 1.5 max gross SR in sample, unresidualized, unconstrained and standalone. That would need to be blended with a hundred un correlated others to have a decent portfolio.
What does unresidualized mean? What do the need to do to the software to “residualize” it? If that’s even the right expression.
It means to remove a set of factors from the signal - or the returns of the signal, which would impact its performance or exposure, and are not meant to be captured by the signal.
Imagine a "I have no preference" signal which gives a weight of 1/n to each asset of the portfolio - you have 100 stocks in your basket, you give 1% to each. If you would compute its performance, you would see that this portfolio would roughly make 10% average returns over 20 years. That seems good right?
Well no it's not, because these 10% don't come from anything meaningful that your signal did. It came because you gained exposure to the market, and that exposure "carried" your signal performance.
> What do the need to do to the software to “residualize” it?
Well residualizing overall just means that you want to regress your signals on a set of factors that would carry its performance for no justified reason. In practical terms, you could start by doing a linear regression of your signal against such factors (say beta, country, sector) and only keeping the remaining residuals (alpha).
Re: FinRL: The first open-source project for financial reinforcement learning
#9Re: FinRL: The first open-source project for financial reinforcement learning
#10As someone with 10+ years of exp in the industry, I had no clue what this was actually doing after reading the github page. Had to dive through the examples and videos. The alphas showcased seem weak: 1.5 max gross SR in sample, unresidualized, unconstrained and standalone. That would need to be blended with a hundred un correlated others to have a decent portfolio.
What does unresidualized mean? What do the need to do to the software to “residualize” it? If that’s even the right expression.
It comes from the “residual” of a linear regression.
So you take some obvious variable like the market basket of stocks, do a regression, and then do your machine learning on the residuals.
The reason is that you don’t get “credit” for doing whatever the market did, you only get credit for doing even better.