How do you handle multicollinearity and overfitting when building predictive models on financial time-series data?
