How do you address the issue of overfitting when training a machine learning model on historical financial time-series data?
