Pandas - how to use groupby with a rolling datetime window to calculate trailing mean with a lag?

Question

I\'m trying to calculate rolling means within a 1-year lookback window, not including the current observation in the mean. But the lookback window does start w