Not to much time to comment right now on specific pokes, but I've got a suggestion for the stats. Often when looking at common EV sets you see a lot of 4/252/0/0/0/252 type spreads, with three different "sets", one for each position of the 4 EVs. These 4 EVs are not significant, and often one or more of these "sets" is lost in the other, or some other set is kicked into other to make room for the three clones. Would it be possible to ignore the difference between 0 and 4 EVs when putting sets together to get a better picture? (For example, all of those sets would just show as 0/252/0/0/0/252, in one category)
This would apply to common 252/0/252/0/0/0 type walls too, which would further compress the stats.
I'm already doing *some* combining--namely, if two EV spreads yield the same stats, they're combined. This is really only an issue for Little Cup, where people don't always realize that if you put 252 EVs in most stats, you're wasting a shitton of EVs, but it does mean that if you do things like 4/253/0/0/0/253, that's not marked as a separate stat.
I probably won't do any more than that, though: I've heard quite a few debates about where it's best to put the "extra four."
would it be easy for you to get what the stats would have been without the weights? it would be interestign to see how much is shift how much is just the weighting.
The data's all there in the "Raw" column, so you *could* compare if you felt so inclined.