Question about MM algorithm: Riot, what do you assume is the impact of one player on the match?

Mephisto3·1/13/2018, 7:45:02 PM·1 votes·382 views

Hi Riot - I know you prefer not to talk about this, which I understand generally.

Perhaps the following question can still be answered: what impact does one player have on the game (as in "what does your algorithm to make fair teams assume")? Do you have some sort of "estimate" based on the data that is available to you? Do you assume the same positive and negative impact?

Would really help for our project.

If too confidential, understand as well.

M3

9 Comments

Mysticman891/13/2018, 7:51:46 PM2 votes

They've mentioned previously that they're only looking at MMR when matchmaking, and don't look at streaks or individual game performances.

So essentially someones impact is proportional to their mmr, and they try to have similar MMRs on either team. Those who have a greater than expected impact for that mmr will rise/fall accordingly after they've played enough games.