Early game champion / late game champion analysis

I always hear about players saying "Champion X has great late game", or "Champion Y falls off really hard". But I wonder how much of the players' perception can be proven using actual game data?

I thought of one idea. Suppose we think, say, Vayne is very strong late game and very weak early game. Then we should expect Vayne to lose a lot more often if the game is short and win a lot more if the game is long. To do this, I drew 860k ranked solo queue games played in Patch 4.6 and computed each champion's win rate segmented by game length. The result is summarized below.

For example, we see that Vayne wins around 45.73% of the time overall. If the game is less than 25 minutes long, however, she only wins 41.31% of the time. On the other hand, for games over 50 minutes, Vayne wins 52.65% of the time. That's a huge difference!

Some of my personal analysis are below this table.

Here's what I can see:

- Karthus, Twitch, Vayne, Veigar are late game monsters.
- Kennen, Shaco, and Alistar are surprisingly strong late game.
- Anivia and Ziggs seem to have WEAK late game, which is very interesting.
- Jinx's late game does not seem to be that strong. She seems to be more of an early game champion.
- If you look at Tristana, it seems that she wins a lot more in very short and very long games, while lackluster in a medium length game.