This is a quantitative traiding project in which is implementated machine learning, genetic algorithms and mathematical indicators techniques for automatically buying and selling positions given specific data about an stock in wich the trading-bot should operate with for generating profits.
In the py_code directory you'll find in what im working on. The file where the indicators operations are located is the stocksOps file. This file helps to threat the data used in the modesl and to calcute the most importante financial indicators.
Recently I added a custom pythorch Deep Q Learning network that "works" and returns indiscrete actions. The network is not working that great. Can't improve returns more than 1%. Maybe work in the alrotihm. The network its kind of very basic, so i'll be working on more complex version.
The Neat Bot network demonstrates improved results, achieving returns ranging between 20-30%.
- Develop NEAT algorithm/network in pythorch.