Mid-lane strength over time

tl;dr Mid: Annie early, Lux Late

Today I am continuing the series on champion power over time by looking the performance of mid-lane champions as games progress. The plots in this post draw data from 840,730 ranked NA solo-queue matches played on patches 5.1-5.7 by players of all skill levels.

Similar to last week's article about marksmen, these plots are made by looking at all the games where a given champion is picked for mid-lane (so there is no Annie as support data here, for example), and then binning those games in two-minute increments from 20 to 60 minutes, and then calculating that champion's 'advantage', for each bin. The advantage is simply 1 if teams with that champion won every game in that bin, -1 if they lost every game, and interpolated between otherwise.

Disclaimer: League of Legends is a team game; I am not at all implying that having a certain champion on your team at a certain point in a certain game is a good or bad thing. There is a mind-boggling plenitude of conflating factors when trying to make plots like these (which is basically why I love this game so much), and so these insights must be considered for entertainment purposes only.

Those with giant statistics egos prepare to have your jimmies rustled, and the rest of us can have some fun thinking about mid-lane champion power at different points in the game.

Here are some plots for champions that seem to win games that end quicker:

Annie, Ahri, and Malzahar all have graphs with roughly similar shape. To me, that is neat, because they have many similarities as champions. They are all bursty, squishy mages, and they all have hard crowd control. Note that the y-axes on these graphs varies, so we see that over this dataset Annie and Malzahar have a bigger advantage early than Ahri does.

Also, and sadly for an analysis like this, not very many games go super-late. Many of the data points on the far right of these graphs are based on too-small sample-sizes to draw many conclusions from. For instance the far-right bar on the Ahri plot is only based on data from 215 games.

Regardless, these plots are higher earlier.

Katarina's bar at 20 minutes is by far the highest of those seen in this post. She has the definite potential to get fed and induce forfeits from enemy teams. However, her plot does trend down pretty sharply over time.

Talon also sports a downward trend. Notably, that strange outlier at 56 minutes on his plot is only based on a pathetic 87 data points and so can not be trusted. His plot remains above zero for the most part, so he must have a decent win rate. It still looks like it is a good idea to try to win with him sooner than later.

The shape of Xerath's plot surprised me. LolKing currently lists his win rate at just below 50%, and that is corroborated by most of his bars being below zero here. However, I expected him to do better later. In my personal experience he is a force to be reckoned with if a game goes late and his team is able to siege up. Also, his wave-clear is so good, it is easy for him to draw games out and get to that point. Anyway, he is not as good late as I thought he would be.

Here are three plots that peak in the middle:

Kassadin has the clearest mid (35-45 minutes) spike on any of these graphs. This agrees with my experience with the champion. He definitely feels strongest to me at that point.

Look at the y-axis of Twisted Fate's plot. He never has any notable advantage by this measure. The last data point there for him is only sourced from 44 games, and so should be ignored.

Viktor looks to be weak before 30 minutes, and perhaps less weak later. All of his bars are relatively small as well. There are champions (we will see them later) that have much more distinct late spikes, though.

These two champions did not fit into the simplified conception of analyzing early/mid/late game:

Both Lissandra and Nidalee seem to have pretty flat graphs. Nidalee is like Twisted Fate above, the advantage numbers are super-tiny in magnitude, indicating that she does not have any particular strength in any of the early, mid, or late game. Unlike most of the other plots in this post, the super-late spike of Lissandra may have some merit to it. There is close to 1,000 games of data in those last 4 bars (though, the last one is only 67 games). I have not played in enough games with her to say anything definitive.

Here are the late-spiking champions:

This plot is hilarious. You either get intimidated by LeBlanc, and forfeit to her at 20 minutes; or you hang on for five or ten more minutes and smash her in the mid game; or you wait around for her to scale up and assassinate your entire team.

Lux and Oriana have similarly shaped graphs, as you can see. They are both squishy mages with damage that scales well with items, and they both have unreliable crowd control. Oriana's bar at 20 minutes is also kind of funny. I can just hear the chorus of annoying teammates crooning 'this Ori sux, ff@20'. If you do that, then you are almost certainly throwing away some wins.

Zed and Yasuo are also similar champions with similar plots. They are both melee AD assassins that struggle to farm early against popular ranged mid mage picks but scale to scary heights in the late game. Either can be used to split-push or team fight to good effect as games drag on. Though neither has as-nice a plot as we saw for Vayne last week, they are both late-game power houses for sure.

Like divers in the briny deep, we find ourselves at the bottom again—having seen so much. I do appreciate you being here with me today; as the kids say, you are the real MVP.

This post, to me, is a good one. Sorry if your favorite mid-lane pick is not represented here. Interestingly, the dataset does contain at least one game for each champion picked as a solo mid-laner. I chose these 16 champions because they are the ones I have the most data for in this dataset. If you want to see the plot for another champion, or you have any thought to share, feel free to get at me on twitter.

I will be back next week, probably with a similar analysis of popular top-lane picks. In the mean time, head back to the index and make sure you have read every single article. Then hit up the rss feed and find out within seconds of next week's post going up.

Beautiful springtime air here (40deg N lat), see you next week for more League and Math insights here at LeagueMath.com.

No late-game Xerath turtle strats were harmed in the making of this article.

Peace.