The problem with using winrates, especially for Champions people tend to main

http://i.imgur.com/ydfPVUA.png

People have been talking a lot about winrates recently. Veigar's winrate is down drastically, for instance. While I agree that Veigar was absolutely wrecked by the latest nerfs, I wanted to say: we shouldn't rely too much on winrates, especially for Champions (like Veigar) who are usually played by Summoners who main them.

Why? Simply put, these are matchmade games. If you win a match, it gives you a harder subsequent match. If you lose a match, your next match is easier. If I played AP Garen Support for 100 matches, my AP Garen Support winrate would be about 50% by the end of it, because the system would adapt and tank my MMR until I was playing against people whose best Champions and playstyles were only about as strong as my AP Garen Support. It works the other way, too: let's say Riot releases Pingu, an absurdly OP Champion. If I play 100 matches with Pingu, I'll have about a 50% winrate by the end. Even if I had an 80% winrate with him to start with, the more matches I play, the more my games will be telling the system, "Pingu has a 50% winrate".

Lest anyone think I'm saying that Vegiar wasn't nerfed, I assure you, that's not my point. Veigar's nerfs are absolutely ridiculous. It is now no longer possible to stun a target unless they're already CC'd or they choose to run into the wall for some reason. His rotation gets minion-blocked. He's completely broken. They can buff his raw numbers to the point where he'll be viable, but his gameplay has become extremely flat: no more 1v1, and your teamfight is reduced to "walk over, throw your stun somewhere, lob damage indiscriminately until you die." That said, because people tend to main Veigar, I think his winrate will start rising back to 50%.

But even if Veigar's winrate starts climbing back up, that shouldn't be taken as evidence that "people are adapting". It's much more likely that it's the system adapting, at least to the extent that my theory is true that most Veigar players are Veigar mains who play him almost exclusively. (Speaking of, there's another winrate problem: if you play a trash Champion almost exclusively, you'll be giving major winrate buffs to any Champion whom you also play.)

A better metric would have to take into account the MMR at which those wins and losses happen, and should probably also factor in total number of games played: if you're at 50% winrate at 1500 MMR with your main, and then Pingu comes out and you play him, and right out of the box you have even a 50% winrate with Pingu, that should be evidence that Pingu is drastically OP, because you're playing as well with him as you were your main, into whom you'd invested thousands of games of practice.

EDIT: Actually, I just realized Riot now has a great dataset for this: they have Champion-specific MMRs from Team Builder. If you want to know how badly nerfs hit Pingu, you should look at how Summoners' Pingu-specific MMRs have changed since the nerfs. Perhaps compute a ratio of Pingu-specific MMR to overall Normal MMR, or to a "basket of (whatever role Pingu is)".