9.8. Trading on stock return predictions

The Lehigh Machine Learning Stock Prediction application was built to explain how practitioners create models to predict stock returns, turn those predictions into a tradable portfolio, and assess its performance.

The app

  • Uses well-known predictive features from academic finance research downloaded from Open Asset Pricing

  • Builds stock prediction models using estimators from the simple (OLS) to the advanced (Neural Nets)

  • Evaluates how profitable trading on these strategies are

  • Contains a practical explanation of the methods, the data used, and how it’s done

  • Compares different models to see which makes a portfolio that generates the highest returns

  • Digs into the details of each model to assess risk-adjusted returns and more

Visit the app and check it out!