Tools: Python, Qt, Cython, scikit-learn, Keras, numpy, pandas, matplotlib, JupyterLab
Quantitative (algorithmic) trading has been something that has interested me for quite a while. I see it as a challenge, as a game of sort, where the rules are barely defined and the space for creativity and opportunity is endless, and yet seemingly unreachable. So in 2017 I have finally started to spend some time on it and it has been quite a trip!
Definitely the hardest thing I’ve done in my life, mostly from the personal standpoint. Embarking on a project like this is a business problem and when you are doing it without a structure/organization around that provides external motivation and direction, it serves as a giant mirror to your own strengths and weaknesses. It forced me to look at myself honestly and address some weaknesses that normally get masked out or muted while working in bigger groups. In a group you can only deviate so much before the group pulls you back. So it’s been a huge personal growth experience, a very sobering and humbling one.
Custom charting and backtesting platform
As a part of this process I have also developed a custom charting tool and a backtesting engine to address some of the shortcomings of the existing solutions. Namely, this one allows me to:
- analyze and backtest in parallel across multiple processes
- draw hundreds of charts at the same time
- smoothly scroll through time and multiple timeframes
- visualize custom time-related data
- visualize backtested trades
- easily label data for machine learning
- visualize labelled and trained data