Better System Trader

192: “Predicting profitability using machine learning” – Ernie Chan



Quant trader Ernie Chan from joins us to discuss how to predict the profitability of trades using machine learning, including: Unconditional probability and the problem with win% in backtest reports, Why “conditional probability” is much more useful for a trader and how to apply conditional probabilities to capital allocation, Why you should use Machine Learning for risk management and capital allocation only (and not for building trading strategies), Why feature selection and feature importance ranking are so valuable, Plus, the best ML techniques for prediction, how simple should trading strategies be, insights from cluster-based feature selection, do these techniques really work, ML as a crystal ball and much more!