HFT Traders Struggle with Shifting Market Dynamics

Most quantitative traders treat Machine Learning (ML) like a 'magic box' that can be fed historical data to produce a signal. However, this approach has a fatal flaw in the world of High-Frequency Trading (HFT) and Market Microstructure.

The market is constantly changing, making it impossible to build a model that assumes the statistical rules of the game remain the same. In HFT, the rules change every millisecond, rendering traditional ML models ineffective.

Standard ML models are built on the assumption of non-stationarity, but in practice, this assumption is often broken. Quants need to adapt their approach to the rapidly changing market conditions, but this is a significant challenge.