Utilizing rolling() with an apply() function with groups in pd.DataFrame object

Question

I am trying to calculate the gradient between each row of the same group with

def get_gradient(row, row2):
    return (row[\'foo\'] - row2[\'foo\'])/(row[\'ba