Is the match making system really that accurate?

12345 ... 8
Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-20-2012

I am not so sure that the elo system matches fairly. I am not the best player by far, but I do try to evaluate how i could have done better in the games I won or lost.

Currently I am floating between 1220-1270 elo at about a 50% win ratio. With that kind of win percentage it would seem I am place correctly.

What I have noticed from looking up the players who I am qued with or against is why I feel like I am being unfairly placed. I find I am usually last pick. This means I am the lowest elo in que, which is fine, but when I look up the players I have been paired with I have noticed they usually fall in the 1330-1460 elo range.

This seems like a good thing for me if I win, but if I don't then I am losing more of my already low elo playing with these higher skill capped players. I personally feel that this 100-200 match-up elo range is too great. I would prefer facing players closer to my current elo and climbing the ladder gradually.

I have read a lot of QQ threads about "elo hell". I am not trying to complain here or blame my teammates. All I am saying is that from my own experience in solo que, I have accepted I have to take the good with the bad, do my best, and try to stay positive. The thing that bothers me is that being matched with higher elo players frequently makes me feel that I should be close to or have the same elo as they do, but because my actual elo is far from the elo I am qued with I fear I wont be able to climb higher.

Perhaps I am just in limbo meaning I am better then the 1200 elo range but not quite 1300, but I would still prefer being placed playing with players in the 1200 elo bracket until I have enough experience to break the 1300 mark.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

In your 10 most recent matches, I see three ranked matches.

Taric, Nunu, and Alistar.

Which of these was a match which you looked up the other players to see that they were all in the 1330-1460 range? I want to look for myself and see the exact Elo of all 10 players.

Also, are you solo-queuing or duo-queuing?


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-20-2012

Taric and Nunu were both solo que. Since I am usually last pick, I just try to fill a role. I find normally teams don't like to play support or a tank, so I try making up for both hence Nunu and Taric. Alistar was a twisted treeline game. I just looked up the players in client after the match had ended.

I did duo these two games with a friend, but I find the same is similar when I solo que. I just don't have a recent one in my history.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

Duo-queuing has an impact on match-making.

Without a game in your recent match history, it's hard for me to really look and try to even guess at what's going on, so perhaps bump this thread when you have a more recent match?


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-20-2012

Alright, I just played a solo. The elo range in this match up is still pretty broad. The lowest elo player was 1130 and the highest was 1355 (looking at both teams). Which is still roughly a difference of 200 elo. And, as they should, the higher elo players played better in general.

I lost 17 elo on this match (which feels like a lot of elo to me) putting me at 1205 although I was one of the lowest elo players in this particular game.

My concern isn't so much who won or who lost. It is more about the range of elo that get paired with one another. If I am one of the lower elo persons in my solo que match ups, then due to order picking, I am usually unable to be put into a position (solo lane, jung, etc) to try to help carry a game. But if everyone in this match-up was within 50-70 elo from the highest to lowest I feel the game would have played out more equally win or lose.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

You're talking about your loss as Taric, correct?

Okay...

Taric - 1272
Sivir - 1130
Lee Sin - 1323
Akali - 1355
Teemo - 1384
average - 1288.8

VERSUS

Leona - 1286
Shyvana - 1283
Morgana - 1349
Miss Fortune - 1181
Shen - 1234
average - 1266.6


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

Okay, that post was just to get the facts down before we analyze it too much.

Anyway, the reason you lost 17 points (as opposed to fewer points) is because with the higher average Elo, your team is essentially a projected favorite. It's like going to Vegas to bet on a football game. You have to bet more money to earn the same profit if you're betting on the favorite. If you were on the team with the lower average Elo, you would have lost fewer Elo for losing (as your team is projected to lose based on Elo).

So that explains that. Now to address your other issue...

In this match, you were 4th on your team in Elo, so you should have been 4th pick, right? Here's what I want you to do... keep a very, very good record of all your ranked matches. If we think something is happening more often than it should, our brain has a tendency to think it's happening more frequently than it actually is happening. So, keep a record. Play like 20 matches, and at champion select on all these matches (you have to also include champion selects that get dodged by someone) write down the pick position you were in.


Now, I will agree with you that the gap between 1130 Sivir and 1384 Akali is larger than I think you should be seeing in ranked queue, this could potentially be explained by duo-queuing. It can be hard to tell who is duo-queuing on either team, and that can have a drastic impact on match-making.


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-20-2012

I didn't realize it went off team average. Now I know my rating wasn't 1272 going into that game. It was around 1220. Which means that my rating shown isn't my actual rating. This must hold true with the other players as well. I am assuming the match makeup is determined by actual not top rating, correct?

Maybe all of the players in this match were around the 1200's in actual rating and I misunderstood what I was reading.

The team average match up seems like a strange way to do things to me. You could get that same average with a whole range of players from different elos.

For instance:

1100+1100+1300+1200+1600= an average of 1260 elo for that team. But the 1600 elo player should be able to carry in this game for his team or might be held back by the players that are 300-500 elo lower.

This would explain why I have been paired (rarely) with players over 1500 elo. I understand that the math shows the team is equal, but is it really? This might be the reason a lot of people rage. Not because they are paired with bad players or "noobs" as much as they are simply paired with players on a different skill lvl (lower or higher).

Again, just hypothesizing and investigating to try to understand why so many people QQ about an elo hell or the real reason people may be stuck in a particular elo.


Comment below rating threshold, click here to show it.

roskosp33

Senior Member

12-20-2012

I appreciate your responses. I can keep track of my next 20 or so games. Do you suggest just que position or should I look for more then just that? Who I am being qued with for example.


Comment below rating threshold, click here to show it.

PogoPogoPogoPogo

Senior Member

12-20-2012

The match-making system is designed for as narrow of an Elo range as possible. When players with large Elo gaps duo-queue together, it throws a bit of a wrench into the system, and that could potentially explain why you have two 1100 players in that particular match.


As for the Elos, I just used everyone's Lolking Elo. If yours isn't accurate, theirs isn't either. I suppose you could go back through and check by searching each summoner name in your LoL client... but they may have played a game or two since then and their Elo has changed some, etc.

As for checking out whether you're highest/lowest in pick selection, I'd just write down what order you are in picks. I'd make a note in any of the matches where you know someone on your team has duo-queued. To my knowledge, most duo-queues either end up 1st and 5th pick or 4th and 5th pick, but almost always, one of them is 5th pick. But again, if you can, try to make note of the pick positions of the duo-queue players, etc.

The whole point of keeping track for a stretch of a somewhat decent number of games is so that we can remove any bias your brain might have toward thinking you get one position more often than others. Ultimately, 20 games is a pretty small sample size, but we should get some sort of a result out of it.


12345 ... 8