Welcome to the Forum Archive!

Years of conversation fill a ton of digital pages, and we've kept all of it accessible to browse or copy over. Whether you're looking for reveal articles for older champions, or the first time that Rammus rolled into an "OK" thread, or anything in between, you can find it here. When you're finished, check out the boards to join in the latest League of Legends discussions.

GO TO BOARDS


LOL Matchmaking Explained

Comment below rating threshold, click here to show it.

Zileas

VP of Game Design

09-16-2009

Quote:
Sayan:
Match making works for me and that is all I care about, I find myself wining more games when I solo que though...

Anyways Zileas I am going to borrow your post and post it in another forum that has been blatantly saying MM in LOL is stupid so it might help explain a few things to them...

Thanx for the detailed explanation, much appreciated...


Hah, what forum?


Comment below rating threshold, click here to show it.

Savage

Senior Member

09-16-2009

Quote:
Zileas:
Oh, you still are seeing on non-peak times those guys in the 16501750 range getting a decent # of lopsided matches. But it's closed beta, so the pool is small, I'm not concerned.


Well it was a hypothetical, I wasn't thinking of a particular rating range or a specific example. Assuming a non-'average' rating of X, are the potential matchable opponents in the range X-Y to X+Y, symmetrical around that player's rating X; or are the potential first pass matches X-Y to X+Z, where Y != Z, an asymmetrical rating window for matching?

I don't understand how it is possible to deliver both 50% win/loss rates over time and proper ELO adjustments for premades without an asymmetric rating window.


Quote:
Zileas:
Because usually the only valid match ends up being a 5 man premade at ~2000 rating vs whatever we can cobble together between solos and partial pre-mades in the 1700-2000 rating. And the decision is either, wait for another 20 minutes, or just go for it.


Well yes, you go for it in that case, I'd never argue against that.

But in a more generic case, you get, for instance, a 10 man pool that you pull a match from with a 1350, a 1300, and 8 1325 rated players.

My understanding is that in this case, the 1350 rated player will *always* be teamed up with the 1300 rated player, to try and balance the sides for a 50% winning percentage, or as close as possible for the pool. The is the tank effect I'm concerned about, and it cannot affect a 5 man premade.

Why do you consider that preferable to simply randomizing the two teams, particularly as the deviations in rating approach 0?


Comment below rating threshold, click here to show it.

Sayan

Recruiter

09-16-2009

Playdota forums


Comment below rating threshold, click here to show it.

Larias

Adjudicator

09-16-2009

Damnit Zileas, why are you not responding to AIMs! I see you posting!


Comment below rating threshold, click here to show it.

Sayan

Recruiter

09-16-2009

Quote:
Larias:
Damnit Zileas, why are you not responding to AIMs! I see you posting!

Cause he don't like no Mudkipz:P


Comment below rating threshold, click here to show it.

Axtree

Senior Member

09-16-2009

This explanation should be in a section like Advanced FAQ on the main LoL website when it launches.


Comment below rating threshold, click here to show it.

Zileas

VP of Game Design

09-16-2009

Quote:
Savage:
Well it was a hypothetical, I wasn't thinking of a particular rating range or a specific example. Assuming a non-'average' rating of X, are the potential matchable opponents in the range X-Y to X+Y, symmetrical around that player's rating X; or are the potential first pass matches X-Y to X+Z, where Y != Z, an asymmetrical rating window for matching?

I don't understand how it is possible to deliver both 50% win/loss rates over time and proper ELO adjustments for premades without an asymmetric rating window.

It's not possible, and yes, if you wanted a perfect queueing system with millions of players, you'd use non-symmetric envelopes. We approximate, and close to 50% is good enough for us. We are not 1 on 1 matching, we are 5 on 5 matching, so we get good results most of the time.


Quote:

Why do you consider that preferable to simply randomizing the two teams, particularly as the deviations in rating approach 0?


We try to make the teams as close to 50/50 as we can. Our present method of assessing how to determine a team's overall strength probably doesn't do that as accurately as some undiscovered method which we will figure out with more data later. But it's the best guess we have. It tends to cause high/low pairing inherently the higher on the rating curve you get, but this tendency is barely measurable to anyone who isnt on the extreme edge of the queue, and in that case, they have rough matches no matter what we do.

So uh, are you a student or professor at the university in your town, or do you work around there?


Comment below rating threshold, click here to show it.

ihatehumans

Adjudicator

09-16-2009

Quote:
Zileas:
Well, everyone had a big opinion on this, often with very inaccurate assumptions, and I thought I'd set it out there so peolpe could educate each other on it.


Now we can haz game balance thread?


Comment below rating threshold, click here to show it.

ihatehumans

Adjudicator

09-16-2009

Quote:
Zileas:
So uh, are you a student or professor at the university in your town, or do you work around there?


Good Will Hunting?


Comment below rating threshold, click here to show it.

Sayan

Recruiter

09-16-2009

Zileas you should really really really really and I mean really sticky this, like for realzors, ya like do it now, Naow!!!