I have always explored the frontier topics of artificial intelligence and I always wonder: how can machines make decisions!
The above question was my one of favorite questions for about 5 years until I found enough time and motivation to use and apply RL to a real-world problem which was financial markets.
About a year ago I started to read multiple papers on the application of RL to financial markets and I discovered several topics like Trading, portfolio optimization, market making, and so on. At that time, I did not know where to start, or how to implement and apply models. So, I continue to read and read then I found out that there are several methods like Proximal Policy Optimization(PPO) that can be very useful in terms of financial markets.
Also, I was deeply involved with Pytorch and I loved it so much. So, I started to Implement some of those models.
The first models started with simple linear layers and I improved them with other layers like LSTM.
After about a year of working on RL, I can say that RL is a very promising and interesting field of work especially when it comes to financial markets. The RL can be very interesting and powerful in most of the market domains.
So, I believe this topic needed to be worked on and it has a wide range of possible approaches for improvements.
I might send some other posts about how to use RL and what was my results and detailed experience with using it.
Good Luck!