Hands-On Machine Learning for Algorithmic Trading

Cross-validation results and ridge coefficient paths

We can now plot the information coefficient obtained for each hyperparameter value and also visualize how the coefficient values evolve as the regularization increases. The results show that we get the highest IC value for a value of λ=10. For this level of regularization, the right-hand panel reveals that the coefficients have been already significantly shrunk compared to the (almost) unconstrained model with λ=10^-5:

Cross-validation results and ridge coefficient paths