How about doing the analysis of missing BOTH the best and worst weeks?
That would be a better analysis than just missing out the worst weeks as the result is always negative and thus the basis to being always invested.