Is the match making system really that accurate?

123456 ... 8
Comment below rating threshold, click here to show it.

Simply Chaotic

Senior Member

12-20-2012

Quote:
Originally Posted by roskosp33 View Post
Is the match making system really that accurate?

Yes.


Comment below rating threshold, click here to show it.

Jamie98s

Senior Member

12-20-2012

matchmaking in this game is terrible, both in ranked and normals. In normals im usually the person with the most wins on my team and the other team usually has a lot more wins than everyone on my team.

and I just played 1 ranked game and got this guy on my team, look at the k/d ratio http://www.lolking.net/summoner/na/22458187#history

thanks matchmaking


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

Jamie98s, you have no idea how match making works.

Who would you rather have on your team, a guy with 1000 wins and 2000 losses, or a guy with 500 wins and 0 losses?


Comment below rating threshold, click here to show it.

Jamie98s

Senior Member

12-20-2012

Quote:
Originally Posted by PogoPogoPogoPogo View Post
Jamie98s, you have no idea how match making works.

Who would you rather have on your team, a guy with 1000 wins and 2000 losses, or a guy with 500 wins and 0 losses?
how about people even with me


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

Quote:
Originally Posted by Jamie98s View Post
how about people even with me
Match-making tries to get 10 people who are all about the same Elo (even in normals, there is a hidden Elo).

But here's the next question... would you rather wait 30 seconds and get a match with decently even teams, even if there are a couple people less experienced and a couple people more experienced, but the gap isn't massive...

or do you want to wait 2 hours for the system to find 10 players that it thinks are identical skill level to your own?


What's more... the system can only estimate a player's average skill level. There's nothing preventing that player from having an abnormally good or abnormally bad game that one match he gets paired with you.


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-22-2012

Alright, I did the 20 solo games like you asked and this is what I found...

Spot in pick order:

1st 11% with 100% win ratio
2cd 5% someone dodged this game
3rd 17% with 33% win ratio
4th 41% with 20% win ratio
5th 26% with 33% win ratio

The average difference in elo between the highest and lowest elo player in these games came out to be 233.

After doing this test it proves that my brain wasn't giving me a bias of playing the bottom positions more often then the others.

This sadly also tells me that support cannot carry solo que games (easily at least). This is unfortunate since I am being qued up as near last pick most of the time, having to fill the role needed, this forces me to rely on my higher elo teammates to carry the game.

If there is an average elo span of 233 among the players, then this also concludes why I am seeing 1400+ players when my elo is 1200ish. But on the same note, if I am first pick then I can easily carry against players in the 1000-1200 elo range (major factor also being able to choose a role that can carry).

From this test, my win ratio would be much higher if I was qued one of the top picks rather then being stuck with a role that is less likely to carry. But since this is not the case, I did the math, and it looks like my win ratio was 46.5% during these games.

So it would seem that how I was feeling isn't so far from the truth of what has been actually happening.

I am curious to hear what you take from this information Pogo. Maybe you see something I may have overlooked.


Comment below rating threshold, click here to show it.

King Kappa

Senior Member

12-22-2012

You're in an awkward spot, the 1200 elo range isn't supposed to be an elo you stay at and I was there for a long time. 1200 elo is where the placement match players are for the most part, the 1400s are the top end or just regular players and you're the bottom end who the system sorts as being in placement matches. Try your best to get 1300 and you'll find that you fit in way better, it'll usually be from 1300-1400. 1200s are really volatile. Good luck!


Comment below rating threshold, click here to show it.

LogicalTautology

Senior Member

12-22-2012

Quote:
Originally Posted by roskosp33 View Post
Alright, I did the 20 solo games like you asked and this is what I found...

Spot in pick order:

1st 11% with 100% win ratio
2cd 5% someone dodged this game
3rd 17% with 33% win ratio
4th 41% with 20% win ratio
5th 26% with 33% win ratio

The average difference in elo between the highest and lowest elo player in these games came out to be 233.

After doing this test it proves that my brain wasn't giving me a bias of playing the bottom positions more often then the others.

This sadly also tells me that support cannot carry solo que games (easily at least). This is unfortunate since I am being qued up as near last pick most of the time, having to fill the role needed, this forces me to rely on my higher elo teammates to carry the game.

If there is an average elo span of 233 among the players, then this also concludes why I am seeing 1400+ players when my elo is 1200ish. But on the same note, if I am first pick then I can easily carry against players in the 1000-1200 elo range (major factor also being able to choose a role that can carry).

From this test, my win ratio would be much higher if I was qued one of the top picks rather then being stuck with a role that is less likely to carry. But since this is not the case, I did the math, and it looks like my win ratio was 46.5% during these games.

So it would seem that how I was feeling isn't so far from the truth of what has been actually happening.

I am curious to hear what you take from this information Pogo. Maybe you see something I may have overlooked.
Solo Queue or Duo?


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-22-2012

Pogo challenged me to do 20 solo games and record my findings.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-22-2012

Quote:
Originally Posted by roskosp33 View Post
1st 11% with 100% win ratio
2cd 5% someone dodged this game
3rd 17% with 33% win ratio
4th 41% with 20% win ratio
5th 26% with 33% win ratio
I actually don't understand the percentages. How many total games is this and how many times did you end up in each position? This isn't 20 games. If it were, each percentages would be a multiple of 5 or 10 exactly.

Is this out of 21 (counting that dodge) and you rounded?

That would mean you were first twice, second once, third three times, fourth eight times, and fifth five times? Is this accurate?


The sample size is still on the small end (even though it's exactly what I asked you to do). I just figured something as simple as pick order would've been a lot more balanced, so I'm surprised.

Here's what I'd like to do roskosp33, if it's not too much to ask. Can you keep recording this data and post an update every 20 games or so? The results will become more and more accurate as you add more data. And if after a sufficiently large sample size of data, if it's still proving your assumption correct, then I'm stumped and we need to get a red in here to perhaps explain what is going on.


With that said, in terms of what seems to be your main concern--Elo, here's a few tips in the meantime. Try requesting a carry role. You're actually 4th pick more often than any other role, and by your data, you're more likely to be pick 1-3 (not any individual, just overall) then you are to be 5th pick. And 5th pick is the only spot where you have absolutely zero say in your role and are forced to support. Yet you had a higher win percentage from 5th pick than from 4th pick.

If you request say ADC or mid from 4th pick, maybe 3rd pick gets mad and decides he's going to pick it anyway--in which case you just pick support or whatever else you would have been stuck with anyway. Or maybe 3rd pick just dodges. And if you're already 4th or 5th pick, if someone else in the lobby dodges, you get a free chance at being a higher pick.



I'm certainly very interested in this thread and willing to have an ongoing conversation about what is going on here as long as you're willing to keep recording the data and sharing it with us.


Quote:
Originally Posted by LogicalTautology View Post
Solo Queue or Duo?
If he's doing what I asked, this data is strictly solo-queue.


123456 ... 8