I think a valuable lesson sometimes I forget too is regardless of the pattern modeled, if there is an outlier amongst the big three ensembles, you can't just toss it, regardless of model biases. The GEFS never really bit on the mega west based block, we assumed it would trend towards the EPS as it usually does, however the opposite ended up being true. When a highly anomalous pattern is modeled, we need to see all three major ensembles on board for a few days for me to believe it.