Brevan Howard Centre for Financial Analysis Trading Challenge is a 12-week remote competition organised in summer 2019. Participants are given encrypted price data from 6 related assets to build a trading robot using reinforcement learning. I got the third place in the competition. The presentation I gave at the prize ceremony can be downloaded at presentation

In the competition I used OpenAI-gym, Tensorflow, Stable-baselines to train Advantage Actor Critic (A2C), an on-policy RL model with curated price features from the given data. Docker was used to containerise the trained model.

As the trading robot used proprietary code provided by Brevan Howard, the code for winning competition is not made directly available on github. Please refer to my github page for other projects applying reinforcement learning in trading.