Machine Learning in Trading
- Developed algorithmic trading strategies for a US-based proprietary trader using the high-level Deep Learning library Keras.
- The tradable universe was Nasdaq and NYSE listed stocks.
- Deep learning was based on tick-by-tick Time & Sales data obtained through the Python API of Interactive Brokers (IB).
- IB is the largest and most technically advanced broker in the US, and also the world.
- Amazon Web Services (AWS) was used for cloud servers as it gave us the lowest latency to Nasdaq and NYSE data centres located in New Jersey.
- Bleeding edge machine learning and data science tools were used in this project.
- For certain parts of the trading system we ditched IB’s official API in favor of a new unofficial API (IB Insync).
- This API leveraged the newly-introduced asynchronous capabilities (AsyncIO) of Python.