Improve performances (vectorize?) pandas.groupby.aggregate

Question

I\'m trying to improve the performances of a pandas.groupby.aggregate operation using a custom aggregating function. I noticed that - correct me if I\'m wrong -