Still under development...

This is a project to construct a trading bot with with [backtrader](https://www.backtrader.com/) framework with multi-leg Pair Trading. It follows Gatev's work on Pair Trading [1] by constructed a customized Backtrader env (/envs/env_gridsearch.py)

Then a series of follow-ups on Reinforcement Learning is experimented, similar rules but on customised gymnasium environment (\envs\env_rl_*)

Are stored in environment.yml

Start from params.py for configurations.

Then trade_gridsearch.ipynb for grid search the most suitable hyperparams.

Lastly, trade_RL_* for reinforcement learning notebooks

[1] Gatev, E., Goetzmann, W. N., & Rouwenhorst, K. G. (2006). Pairs trading: Performance of a relative-value arbitrage rule. The Review of Financial Studies, 19(3), 797-827.