We started the event learning tons about reinforcement learning and my team went through the deeplizard tutorial listed here:

This was very useful to us and after learning all this information we tried to implement this into our code. We tried using neural networks but we had difficulty training our code. Unfortunately, we were unable to get this working so we decided to use a different approach.

We learned the meanings of different indexes and tried to use that understanding to build a solution. We looked specifically at the CCI (Commodity Channel Index ) and RSI (Relative strength index) to try to help the agent predict which of the three options to do.

Overall this experience was definitely a challenge for a team unfamiliar with reinforced learning. My team learned tons just pushing through and overall this experience was worthwhile.

Built With

Share this project: