Season 5: Statistical outliers
tl;dr Play enough games, and almost anything can happen.
Hello #LeagueMathers! Let us start with the good news. Riot has approved LeagueMath.com for a shiny, new, and very fat data API pipe, so the data for today's post is sourced from more matches than ever before; a lot more.
I have been sifting through 250,081 (yes, over 1/4 million) ranked NA solo-queue matches played on patches 5.1, 5.2, 5.3, and 5.4. Now, this is definitely still a hair-thin fraction of all such games played, bit it is many more games than any one of us will play in our lifetime, so... There's that.
I thought it would be fun to start working with this new dataset by looking for outliers. In over a quarter million games, there is bound to be some interesting and funny stuff. All we have to do is look.
Computer scientists love to sort things. Here is the longest match in the dataset:
Over an hour and sixteen minutes. Yowza. It looks like it a was relatively close game too. Jinx did manage 600 cs in defeat, and she and LeBlanc were both 22/14/24, an interesting coincidence.
And here is the shortest match I have found:
Just six minutes, and no kills. It looks like the red team bot lane never really got off the ground. I was thinking this was boring until I noticed the Lucian who never showed up is named VANISHING DRAG0N. That is comedy.
Moving on, I like to look at the damage dealt to champions graph after a game to see who really carried. Perhaps unsurprisingly, the Jinx in the longest game in the dataset also did the most damage to champions:
And if I have been split pushing, I like to look at the total damage dealt graph. This Miss Fortune got damn near a million:
She did manage to win; but that team looks pretty heavy. Interestingly, her gold earned, 36066, is also the most for a single game in the entire dataset. That is called carrying.
Since I like comeback victories, I personally enjoy the mechanic of inhibitors respawning. Here is a wacky game where the red team killed 15 inhibitors:
They had a twelve-thousand gold lead at 40 minutes; but it looks like blue almost came back. What a nightmare. To lose after killing 15 inhibs would be insane.
In a quarter million games, there must have been quite a bloody one. Here is a match with 157 kills:
The game was a total one-sided stomp for the blue team, but I think they were just having fun fighting. I know it looks like fun to me. I also like the image of the angry Sivir selling all her items at the end.
Next I want to look at individual performances in terms of kills, deaths, and assists. Remember, these games are recent, ranked, NA solo-queue matches; not bots, or normals, or ARAM.
Here is Bjergsen with 48 kills in a single match:
That 12 minute Mejai's Soulstealer + Elixir of Sorcery.
Here is a Warmog's Nami with 53 assists:
That is how you win from support. Tank Nami with 3 kills has more gold than her mid and top solos. GG.
Aaaaaaaaaaaaand, here is how you go 0/38 on Lee Sin and still win:
You sir, are why the tribunal exists.
In previous searching, I had found a game where a Kha'Zix had gone on a 29 kill streak. In the new dataset, I can do one better. Here is a match where a Cho'Gath named HopintheRide went 30/0:
Vision is an integral part of the game that is often difficult to measure. Here is a silly match where some nerdy Sona named LunarMoooooon placed 1739 wards:
ggwp, report 9x, thx.
Here is a match where Gabezilla the thresh god did work, placing 95 wards, destroying 32, and buying 26 pinks; and somehow lost.
Keep fighting the good fight. You'll get 'em next time.
As ever, thank you for taking the time to read my posts. Bonus round: Here is one where a Lissandra killed 672 minions and lost. I guess cs isn't everything. Ha!
This was fun for me, I hope it was for you too. Get at me on twitter and and let me know what I missed. Can you find a bigger outlier on any of these axes? I'd like to see it!
We have gone as far as we can today. Head back to the home page and read another article. Then, once you have read them all, get frisky with the rss feed and come back next week. With the new and improved huge datasets, I am excited for the upcoming posts.
We be doin' league and we be doin' math here at LeagueMath.com.
No intentional feeders were harmed in the making of this article (unfortunately).