Splitters

Expanding Window

Uses all data up until the current split date.

Sliding Window

Useful when you want to limit how much long models should look "into the past".

Single train-test split

If you are not convinced of the usefulness of Adaptive ML, feel free to use the classical single train-test split. The downside is that you throw away data that can be used as your test set.