Forecasting and Artificial Intelligence Based Strategies
– Overview of AI techniques
- General paradigm of machine learning.
- Stepwise linear regression, classification and regression trees (CART), neural networks, genetic algorithm, Bayesian networks, and support vector machines (SVM).
– Extended exercise on predicting returns of a portfolio of stocks using stepwise linear regression, CART, neural network, and SVM.
- Constructing the features set: normalizing technical and fundamental variables, and devising candidate entry rules.
- Constructing the dependent variable.
- Tuning the parameters of stepwise linear regression and CART for features selection.
- Interpretation of output.
- Incorporating selected features and trading rules into a complete backtest.
- Out-of-sample testing and cross-validation.
- Tricks to enhance the classifier: Boosting, bagging, and bootstrapping.
- Comparison with results from neural networks and SVM.
Comments are closed.