Kill participation by role

tl;dr elo hell is where teammates fail to assist one another.

I am often struck by the emphasis the LCS broadcasts place on kill participation. This statistic is calculated for a player by taking their kills plus assists and dividing that by their team's total kills. This stat will favor both teams that group up as well as players in dominant lanes of one-dimensional teams (e.g., marksman and support when bot-lane gets all the kills).

So, I wondered (as I am wont to do), how this statistic varies across both role choice and different leagues (skill levels) in the current meta.

The LeagueMath.com dataset has 1,049,459 recent ranked NA solo-queue matches played on patches 5.3-5.8 by players of all skill levels. By aggregating the kill participation of the millions of players in this dataset we can get a birds-eye view of how this statistic varies in solo-queue.

Here is the table for solo top laners:

Before I did this analysis, I predicted that kill participation would go up with league (skill), regardless of role. Happily, we see that here. Interestingly, the sharpest increase seems to be as players improve from Bronze -> Silver -> Gold, and then things seem to level off.

Top laners have the lowest kill participation in this analysis. This makes sense, because in the current meta top lane is a bit of an island where intense 1v1 battles (occasionally interrupted by junglers) give shape to the first 20 or so minutes of the game. Since many of these kills are unassisted, top laner kill participation will suffer during the laning phase.

Here is the table for junglers:

Junglers have better kill participation than top laners, which makes sense, since they are always moving around the map and so will have more opportunity to initiate and participate in kills. The curve appears to be about the same, with Bronze having much lower kill participation than other leagues with the curve flattening out after Gold league.

Since kill participation does seem to go up with player skill, I wonder if even higher kill participation would be even better? There is the obvious benefit of extra assist gold, but there are tradeoffs to having everyone grouped for every kill. If a team does not spread themselves out over the map, a lot of experience can get lost.

Here is the table for mid laners:

Mid lane kill participation is between that of top lane and jungle in every league. This is believable given how the game is typically played. What is perhaps interesting here is that the drop-off in kill participation for low-ranking mid laners is not as dramatic as what was seen in top and jungle. The best mid laners do not participate in that many more kills than the worst.

In my experience, mid lane is often decided either by jungle and support roaming pressure, or a bad match-up, or one player last hitting creeps much better than the other. Of these three, only jungle pressure has any effect on kill participation, and so we can perhaps rationalize that mid-lane skill will not be well predicted by this particular statistic.

Here is the table for the role of marksman (duo bottom carry):

This table is surprisingly close to that of mid lane. I had to make sure I did not accidentally copy/paste incorrectly. Is it possible that the kill participation numbers for the high ranking leagues in this analysis are at some kind of optimal level? If so, then it may be mistaken to laud players for having super-high kill participation.

Regardless, it does seem that Bronze-level players get a lot more solo kills than players in other leagues. Perhaps there is a lesson to be learned here. I will leave it for you to decide.

Finally, here is the table for bottom lane supports:

Dedicated supports have the highest kill participation of any role, and this is consistent across leagues. Also, the difference between high-ranking supports and low-ranking supports is bigger, and the curve is smoother, than for other roles. Maybe Aphromoo was wrong when he indicated that 'support is so easy'. The design of many of the popular support champions does encourage getting a lot of assists.

Assists are one of the few areas of the game where gold appears out of nowhere. An un-assisted kill is actually quite wasteful, since the bonus gold that any assisting teammates would receive is lost. The amount of gold given for an assist is 50% of the kill plus an additional bonus between 30-60 if the player getting the assist has more assists than kills. This can make a huge difference in gold-earned for the team over the course of an entire game. Especially in solo-queue where games typically have high numbers of kills.

Welcome to the bottom of the post. That was fun and easy (and perhaps even insightful). I would love to do a similar analysis on the data from LCS matches. Competitive games have a very different shape than the average solo-queue game; so it would be interesting to see how these kill participation numbers vary. Unfortunately, the Riot API does not provide the data from competitive games yet.

I will now present the three-step plan to success in succinct list-form:

1. Hit me up on twitter and let me know what you think of these articles.
2. Use the rss feed to get real-time notification of future posts.
3. Go back to the home page and read till your heart's content.

Until next week, keep doing League and keep doing math—that's what we'll be doing here at LeagueMath.com.

No brainless AFK split-pushing bronze-league top-laners were harmed in the making of this article.

Peace.