Pandas rolling() function gives minor different result on two identical dataframes

Question

I have two large pandas dataframes.

One contains 100k rows and second one contains 10k rows.

The last 1k rows are 100% same.

But when I do a operation l