Who wins common lane matchups?
Patch 5.13
tl;dr Solo lanes get the biggest earliest leads, but the game is not decided in the first ten minutes.
I have been playing top in many recent ranked matches, and it got me thinking. How much do individual role vs. role matchups actually matter? Does any specific matchup lead to reliable early disparities in XP or gold? Do those disparities lead reliably to victory or defeat?
For this week's post I looked at the most commonly played matchups, per role, in ranked NA solo-queue matches. In my dataset of 2,265,357 matches from season 5 specific pairs of champions see each other in certain roles tens of thousands of times. By looking at the average outcome of those specific matchups we can hopefully begin to get at those questions above.
Specifically, I measured (for each matchup) the overall win rate, and the average difference in XP and Gold at 10 minutes. The idea is to see not only who wins in the end, but if there are any standout differences in the early game as well.
Here is a table for the most commonly seen top-lane matchups:
| Matchup | Winner | XP @ 10min | Gold @ 10min |
|---|---|---|---|
| vs. | 50.3% | 120 | 148 |
| vs. | 51.2% | 215 | 313 |
| vs. | 51.7% | 31 | 125 |
| vs. | 51.4% | 15 | 131 |
| vs. | 53.9% | 142 | 232 |
| vs. | 54.6% | 109 | 177 |
| vs. | 56.2% | 194 | 210 |
| vs. | 50.5% | 175 | 141 |
| vs. | 51.5% | 173 | 163 |
| vs. | 52.3% | 64 | 11 |
The most common top-lane matchup is Irelia vs. Riven, seen 11,215 times in the dataset. Irelia wins 50.3% of the time, which is mind-bogglingly close to 50%. It looks like Riven has a small XP and gold lead at 10 minutes, but those small leads do not guarantee wins.
The biggest end of game win-rate disparity in the top lane is Riven beating Vladimir 56.2% of the time. She also typically has an XP and gold advantage at ten minutes in that matchup. She gets a bigger lead at ten minutes against Nasus, but in the long run Nasus wins. This agrees with what we know about Nasus being stronger later in games.
Overall, the differences seen in this table are smaller than I would have predicted. League of Legends, at least when played within the prevailing metagame, is extremely well balanced. I wonder when this will stop surprising me.
Here is the table for Junglers:
| Matchup | Winner | XP @ 10min | Gold @ 10min |
|---|---|---|---|
| vs. | 54.5% | 90 | 51 |
| vs. | 54.5% | 105 | 27 |
| vs. | 53.4% | 76 | 19 |
| vs. | 57.3% | 158 | 70 |
| vs. | 51.0% | 7 | 68 |
| vs. | 52.3% | 64 | 110 |
| vs. | 52.2% | 150 | 64 |
| vs. | 57.8% | 54 | 119 |
| vs. | 51.7% | 180 | 164 |
| vs. | 56.7% | 252 | 116 |
This table has some of the biggest win-rate differences seen in any of the roles. Amumu, Sejuani, and Udyr all win a lot against Lee Sin. Welcome to the season 5 tank meta. It is a bit of a surprise not to see Gragas here, but the dataset includes games from patches 5.3-5.13 (approximately evenly distributed) and the rise of Gragas jungle is both a relatively recent phenomenon, and potentially more prevalent in competitive play than in solo-queue.
The Vi vs. Rek'Sai and Amumu vs. Vi rows stand out here as well. Those matchups were seen close to 20,000 times each in the dataset, and on average their XP and gold at ten minutes differed essentially not at all. Of course, in any given game one of them could get a bigger early lead, but over a large number of matches these statistics appear to even out very well.
Here is the table for mid laners:
| Matchup | Winner | XP @ 10min | Gold @ 10min |
|---|---|---|---|
| vs. | 54.2% | 126 | 167 |
| vs. | 53.7% | 15 | 146 |
| vs. | 52.8% | 191 | 301 |
| vs. | 51.1% | 77 | 106 |
| vs. | 51.9% | 36 | 65 |
| vs. | 55.8% | 208 | 124 |
| vs. | 50.1% | 133 | 70 |
| vs. | 56.3% | 162 | 131 |
| vs. | 51.8% | 42 | 256 |
| vs. | 54.6% | 178 | 210 |
Many of these matchups are quite even as well, though it looks like Zed typically gets an early lead in XP and gold. A 300 gold lead at ten minutes corresponds to about 15 cs. There just are not that many waves before ten minutes, so a lead like that indicates that Zed is able to zone Ahri off a lot of early creep kills.
For me, it is interesting to see that LeBlanc typically has an early lead against Annie, but ends up losing a fair bit of the time. I play Annie mid, and this corresponds with my personal experience. I have lost a great many lanes to LeBlanc, but end up contributing more in later team fights. These data include matches from players of all skill levels, so there is significant smearing that would preclude things like super-high-skill LeBlanc skewing the results in her favor.
Here is the table for marksmen (ADC):
| Matchup | Winner | XP @ 10min | Gold @ 10min |
|---|---|---|---|
| vs. | 53.0% | 91 | 123 |
| vs. | 51.6% | 9 | 18 |
| vs. | 53.8% | 92 | 77 |
| vs. | 56.0% | 65 | 81 |
| vs. | 50.5% | 69 | 24 |
| vs. | 51.1% | 34 | 20 |
| vs. | 50.4% | 104 | 148 |
| vs. | 52.7% | 16 | 36 |
| vs. | 54.8% | 37 | 9 |
| vs. | 50.4% | 20 | 18 |
In the first ten minutes of games, these matchups are shockingly even. Single- and double-digit gold leads at ten minutes mean essentially nothing. The biggest typical lead from this table is Graves over Vayne, and with Vayne's win rate of 50.4% in that matchup, we can see precisely what Graves' early lead amounts to.
Jinx is winning many of her matchups despite not reliably being ahead at ten minutes. Personally, I always try to do my best throughout each game, not giving up early. This kind of data reaffirms that attitude for me; in a typical game each matchup will tip a bit in favor of one player or another early, but it is not like those early leads decide the game.
Maybe more interesting than the data in this table is how many times these matchups are played. There is much less variability in the picks for this role than in any of top, jungle, or mid. For instance, the most popular matchup (Jinx vs. Graves) was played 86,393 times in my dataset. And even the tenth most popular matchup (Caitlyn vs. Graves) was played 48,202 times. My dataset is just a tiny fraction of all the matches played, but these same marksman matchups are seen over and over and over.
Not sure how insightful it will be, but for completeness, here is the table for supports:
| Matchup | Winner | XP @ 10min | Gold @ 10min |
|---|---|---|---|
| vs. | 53.3% | 43 | 41 |
| vs. | 52.6% | 34 | 8 |
| vs. | 54.3% | 10 | 21 |
| vs. | 51.5% | 18 | 93 |
| vs. | 50.8% | 36 | 48 |
| vs. | 51.2% | 24 | 143 |
| vs. | 53.3% | 35 | 53 |
| vs. | 51.9% | 19 | 64 |
| vs. | 51.4% | 32 | 12 |
| vs. | 50.8% | 44 | 59 |
Blitzcrank confirmed best hook champion. Just kidding.
These are, perhaps unsurprisingly, the most even matchups of any role. It is hard for a support to reliably gain any kind of significant early advantage.
Morgana's gold leads here probably follow from how easy it is to proc Spelltheif's Edge with her Tormented Soil (W). A small early gold lead is, again, not a big deal. Morgana's contribution to a game is going to come from clutch Dark Bindings and flash Zhoyna ults, not Spellthief's money.
At the end of this post I feel that the battle has been won, but the war rages on. The content of this post would make a fun site. Choose a champion, and see how that champion fares in common lane matchups. Then you could refine the results by filtering matches of players around your skill level, either by typing in your summoner name or choosing your league. I would love to know how Maokai does in common top-lane matchups. Something like http://www.championselect.net/, but based on actual game data. Let me know if you build such a thing.
Have a thought? Let's tweet.
Not done yet? Get at the index.
Next week, another article: Subscribe to the rss feed.
The central limit theorem underlies results like those in this post, and the law of large numbers goeswith. Just as League of Legends goeswith Math here at LeagueMath.com.
No unpopular champion matchups were harmed, or even considered really, in the making of this article.
Peace.
