Going Ever So Slightly Beyond Champion Pools
Today I would like to introduce a new statistic professional League of Legends — effective champion pools or ECPs.
Effective champion pools are a simple extension to the current champion pool statistic that measures how many champions any given player has played. The key difference is that ECPs take into account how frequently a players plays different champions.
Think, for example, about two different players A and B who each play 45 games. Now imagine player A plays the same champion 40 times but for the remaining five games they play five different champions. By the traditional champion pool metric that player has played six champions. Now player B takes a different approach and players three different champions seven times each and another 3 champions 8 times each. By the traditional champion pool metric, that player has also played six champions.
Do these players really have the same champion pool? Not really. This is where ECPs come in handy.
First, I should briefly explain what goes into these values. You can also find the code for this analysis at my Github here.
At its heart, this is simple an application of the Simpson diversity index. The calculation is just the inverse of the sum each champions percentage of play (calculated using the formula 1/sum(p²)).
Returning the previous example of players A and B, we can calculate two separate ECPs. Table 1 shows a dramatic difference between the Champion Pool of the two players and their ECP. While player A has played 6 different champions, they effectively only played 1 and a quarter. That passes the gut check — after all, player A used the same champion for 40 of 45 games.
Player B, however gets credit for nearly six full champions. They likely deserve credit for a full six — the last three games can be chalked up to noise — but 5.97 is very close.
So what can we learn from this statistic? I’m going to focus on North America, looking at data from the 2020 season. To keep things as comparable as possible I am limiting the data to only regular season games. All the data for this article comes courtesy of Oracles Elixir.
Let’s start with an easy one — who played the most champions during regular season play in 2020? It’s a tie between Nisqy and Ruin at 18 champions each. It’s actually marginally more impressive for Ruin than Nisqy because Ruin only player 35 games instead of the 36 that Nisqy played. They’re followed closely by Licorice and Broken Blade at 17 each with 36 games each. Huni also deserves an honourable mention for playing 16 champions over only 28 games.
So who tops our list when we consider effective champion pools instead? Ruin takes the prize at 11.24 effective champions played and Broken Blade follows close behind at 11.17 effective champions. No one else breaks 11 but Jiizuke rounds out the top three at 10.92 effective champions.
The lists are not exceedingly different (and remember that I promised they wouldn’t be). However, the differences are worth exploring. Which players get penalized the most by the use of ECP over raw champion pools?
Froggen is the clear winner on this front. While he played 13 different champions over his 26 games in the LCS, his ECP is only 5.2 — a drop of more than half. That’s largely because he played 10 games of Orianna, an impressive number of games of the same champion.
Other players who take big hits are Nisqy (18 compared to 10.4 ECP) and Licorice (17 compared to 10.0 ECP). They played 8 games of Zoe and Sett respectively which likely contributes to this drop although both did double up on a bunch of other champions as well.
It seems like a pattern with discussions of analytics in League of Legends, but its still important to say that this statistic only provides a small part of the picture and everything is context dependent. When Doublelift famously played Lucian practically every chance he got and when asked why, he said because it was the best champion. A lack of diversity in a players champion pool can indicate they were never forced into playing anything else as much as it can indicate they don’t know how to play anything else.
As the title of this article suggests, this wont revolutionize professional League of Legends. I hope only that it can add to the depth of analysis and that someone somewhere will take this and run with it.
