League of Legends champion pairings
tl;dr Fiddle / Talon, the new meta.
Like the right wine with dinner, or matching socks, success in life often comes down to how things get paired up. A lot of sites look at how well particular champions do on the Rift, but I have not come across an analysis of how well pairs of champions perform. So I made one.
Today's dataset: Over sixty-four thousand NA ranked solo queue matches played by diamond, master, and challenger elo (MMR) players on patches 5.1, 5.2, and 5.3. There are 7,503 possible pairings of champions (123 choose 2), and a good fraction of those have actually been played in ranked matches. By looking at the most basic metric, win percentage, we can get an initial idea of how well each pair performs.
This first table shows, of the pairs that were played at least a hundred times in the dataset, which had the highest win percentage:
Fiddlesticks + Talon, the new meta. Ha! Okay, maybe not. But honestly, I didn't expect to see any of these pairs. Where is Janna / Graves? Janna does appear in one of the pairings, but not with a marksman, apparently the secret sauce is Singed. I also like seeing Vladamir / Irelia near the top. Both are typically top-lane champions in the current meta, but the combination does sound brutal, does it not?
I am also pleased to see that these win-percentage numbers outstrip the typical values seen for single champions. With the right pairing, the whole can perform better than the either of its parts.
The next table is a counterpart to the first, of the pairings which were played at least a hundred times, which had the worst win percentage:
LeBlanc + Karma does sound a bit weak, but Syndra + Gnar? That sounds like a powerhouse to me. The data does not lie, though; in over 150 high-level ranked solo-queue matches played in the dataset, Syndra / Gnar won only 31% of the time. Ouch. Lux + Lucian sounds bad too me too, but I have never really liked Lux.
Perhaps these pairings should be avoided.
What if we take away the restriction of having played together a hundred times? The next table summarizes who wins the most in that case:
Yes! Olaf + Karthus. I cannot really explain that. Let me know if you have any thoughts. I do like how this list is full of what I might consider to be cheese-ball champs. Like, 'yup, we just got rocked by Nocturne / Darius.' Of course, the sample sizes are much to small to take these results seriously, but maybe there are some hidden team composition ideas in here somewhere.
Hiding near the bottom there, with a respectable win rate of over 83% is Taric + Varus. Essentially the most traditional pairing of support and marksman in this entire analysis. I do not believe I have had the pleasure of facing that particular lane, but it sounds mean as hell. So much lock down.
Now, for fun, here are a few pairs that over the entire sixty-four thousand match dataset never won:
I have no words.
Though, Ziggs / Azir does seem like it should have some merit. Oh wait, there is more to this game than doing tons of magic damage from 2 screens away.
We can finish with the companion table, here are the pairs who have yet to lose:
Rammus + Kha'Zix. Ok.
And there is Eve again. I guess the secret is to pair her with Shaco (yowza, so much invis!) instead of Shyvana. Ha! And what is Yorick doing up here? Ah, the magic of insufficient sample-sizes.
Shyvana / Volibear does sound like a slabby front-line. Ok, that will be quite enough silliness for now.
Yo! Thank you for taking the time to read this. You are the real MVP. I hope these pairings spark some new ideas for you and your duo-queue partner. If you give any of these a go, or you think I have missed something in my analysis, let me know! I love hearing from you guys.
Personally, not many of my favorite champions showed up in these tables. Maybe that explains my recent mediocre results. Ha! All right, I'm off to try out Tryndamere + Nidalee.
No APIs were harmed in the making of this article.