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


Elo is dom.

123
Comment below rating threshold, click here to show it.

kavi

Senior Member

04-18-2013

Quote:
Eloquid:
So, if one is say at +50 wins, does it mean he or she is at 1200 (average of averages, if I remember correctly) + 50*12 = 1800 elo? I don't know, it doesn't sound like high win/loss ratio, yet elo thus calculated is pretty high.


No. Again, initial Dominion elo is based off your normal Elo at the time you start playing Dominion, which one would need to know (or make a very good educated guess). On the other hand, +50 wins is actually quite a lot, and a person with that quantity who started at an average Elo should have an Elo of at least 1600, and possibly 1800 or higher. The top tier on Dominion tend to have very high win-loss difference compared to their Elo due to the fact that they are continually matched against teams of lower MMR and thus gain a lot less Elo per win. But, most players at this level have the ability to win consistently regardless.

Ergo, the correlation of win-loss and Elo is a bit of a bell curve with the best correlation being at average Elo. This is not a phenomenon unique to Dominion, and happens with top-tier Elo players on SR as well. I think it has as much to do with how the matchmaker works as with the lower population base of Dominion. Given that 90% of the playerbase is between 1000-1400 Elo, this is fairly logical.

Quote:
DiscworldDeath:
You can't extrapolate anything by number of wins. I've seen people with about +10 w/l at 2k Elo and at 900. Not all games are worth the same.

You could sort of extrapolate when you used to have a beginning number, but the more time and games that pass from your last recorded information, the less useful it becomes. And if you multi-queue a lot, just forget it, it's sort of true if you only ever solo queue.


^ This. However, the accuracy for people who almost exclusively play solo queue is fairly high if you do have that beginning number - I did quite a bit of testing on that in the first half of 2012. If you are not above 2k Elo on Dominion, you are almost guaranteed to average out to 12 Elo per win playing solo queue.


Quote:
Eloquid:
I always play solo because I perform better that way. Having friend in a team always makes me wanna help them and save them from their own stupidity, which most often ends up with both of us being dead, so...

I left TT map at something around 180/100 ratio, which would put me at +2k elo, if 12 points per win were true on that map. I don't feel that good though.


Normal TT has Elo in common with normal SR, or it did last time I checked (don't ask me why).

Quote:
Koravel34:
W/L ratio doesn't really tell you anything because we don't know your starting point, and that's assuming all the other factors kavi mentioned don't apply to you, (I play mentor games, so I'm sure that #2 applies to me, at least).

If you were playing last June, you can get a very rough estimate by looking at W/L ratio and recorded elo at that time vs your W/L ratio now.

Another issue would be playing against premades that you don't know are premades, as this also affects your MMR differently.

There really is no way to be sure anymore.


afaik win/loss ratio has absolutely no bearing on MMR (and never did). I think you must mean win-loss difference, which is somewhat... different.

edit: also (I know I'm being picky about semantics here, but indulge me ) the only things that effects *your* MMR are your stats (Elo, level, number of wins) and your queue size. If you are matched against a premade, then yes, there is a substantial chance your MMR is different than the average MMR of the opposing team. However, it's not necessarily *lower*, and this should likewise average out over time. Suffice it to say that if *you* are solo queueing and are not above 2k Elo or so, you are likely to average out to 12 Elo per net win.


Comment below rating threshold, click here to show it.

Koravel

Senior Member

04-18-2013

Sorry, I meant this effects your *elo* differently, as far as how much you gain for a win or lose for a loss.

What I'm trying to say is this, I'll use my case for an example.

As of June 2012, whenever they shut down Jabebot's ability to track elo, I was at 1854 elo and a W/L difference of +25. Now I am at a W/L differential of +47, elo unknown.

I can thus say that it is very improbable that my elo has dropped since then, and it is most likely that it has risen. How much exactly is impossible to tell. Thus, like I said, a rough guesstimate.


Comment below rating threshold, click here to show it.

KaWraith

Senior Member

04-18-2013

Quote:
Koravel34:
Sorry, I meant this effects your *elo* differently, as far as how much you gain for a win or lose for a loss.

What I'm trying to say is this, I'll use my case for an example.

As of June 2012, whenever they shut down Jabebot's ability to track elo, I was at 1854 elo and a W/L difference of +25. Now I am at a W/L differential of +47, elo unknown.

I can thus say that it is very improbable that my elo has dropped since then, and it is most likely that it has risen. How much exactly is impossible to tell. Thus, like I said, a rough guesstimate.


^This is the essence of what people mean when they talk about elo in dominion.

My Elo was 2109 when jabebot was shut down, my win rate is still over 50% (53.97% atm), so in general if my average games are +/-12 for wins or losses, its unlikely I'm at a lower Elo now, and very probably higher.

In terms of +wins over losses, I'm at +81 (551 wins, 470 losses).

But, the real way I know I'm in high elo is because I recognize 85% of the players who show up in games I play because I have seen them before on a routine basis, there are 10 revives in virtually every game I play, and I routinely see players from the top 4 DD tournament teams show up in my games.

As for Koravel, he shows up in my games a bunch too. So he is probably in a similar ELO to mine. how much of a difference his +47 makes compared with my +81? Nobody knows. But the "12*+wins+1200 formula" says that my elo is 2172, and his is 1764... I don't think that this is very realistic.

Riot is determined to stamp out ELO/MMR, and move to the "league system", despite the fact that the league system backstops your losses, and functions independently from your actual MMR. So competitive plays will continue to want to know their MMR because it is an actually a measure of how they are competitively ranked by matchmaking. That's why we talk about Elo: it's a meaningful designation of the skill level of the players we compete with in matchmaking.


Comment below rating threshold, click here to show it.

brainlizard

Senior Member

04-18-2013

FWIW, I was at +60 and 2253 as of the last lolmatches data, and I'm almost entirely solo queue.


Comment below rating threshold, click here to show it.

Eloquid

Senior Member

04-19-2013

^ You just opened another can of worms, since "elo" number that lolmatches provide are pretty dubious


Comment below rating threshold, click here to show it.

Sauron

Senior Member

04-19-2013

It is best to just go by win difference now. It can get skewed by premades, but typically being +75 wins is going to indicate high elo. It can be less if you solo queue a lot, or much higher if you premade exclusively. The fastest way to get to high elo has always been solo queue and that remains true today.

A rating given by a third-party calculator is always going to be wrong. These things cannot take into account premades and opponent ratings (you can gain as little as 0 rating from a win, as we found out back when ratings were retrievable via profile data).

The Lolmatches ladder is simply too old to go by. It's interesting to look at, but like konf said, it is only usable for players with stable win differences assuming they haven't quit.


Comment below rating threshold, click here to show it.

kavi

Senior Member

04-19-2013

Quote:
Eloquid:
^ You just opened another can of worms, since "elo" number that lolmatches provide are pretty dubious


I think if LoLMatches' data was incorrect they wouldn't have gone to such lengths to hide it brainlizard's elo makes perfect sense at the time given who he was playing with/against (though it looks like it's fallen a bit since then).

Quote:
Sauron:
It is best to just go by win difference now. It can get skewed by premades, but typically being +75 wins is going to indicate high elo. It can be less if you solo queue a lot, or much higher if you premade exclusively. The fastest way to get to high elo has always been solo queue and that remains true today.


^ tl;dr for my wall(s) of text, thanks


Comment below rating threshold, click here to show it.

Koravel

Senior Member

04-19-2013

Quote:
kavi:
I think if LoLMatches' data was incorrect they wouldn't have gone to such lengths to hide it brainlizard's elo makes perfect sense at the time given who he was playing with/against (though it looks like it's fallen a bit since then).


Actually, whether or not it was correct has no bearing on Riot's decision to hide it. They didn't want normal queue being plagued by people who would look up teammate's elo and harass them or queue dodge if it was low.

Whether or not that elo was the correct one is irrelevant.


Comment below rating threshold, click here to show it.

Sauron

Senior Member

04-19-2013

Lolmatches pulled directly from the champ select data under the name of "odin rating" which was dominion's working project name, and it operated just like ranked elo. A support rep confirmed my rating seperately as well, and it was in line with lolmatches at the time. No idea what it is now.


Comment below rating threshold, click here to show it.

FDru

Senior Member

04-19-2013

Average queue time is probably the only reliable way to guess elo these days. If you play with "tons of Revives omg" it is not because you are high elo. If you are in queue for 40 minutes between games consistently, it is because you are high elo.


123