I would like to have data from master players.
I have enough games on my main so it’s fair to say that “if” there’s actually a mmr system in place, it would be set at this point.
I’m at the highest mmr, but only 54% WR last season (572 games). It’s a bit telling in my perspective. I can also feel that the opponent isn’t as bad as on my smurf.
ignoring what people post doesn’t mean you haven’t been ‘proven’ otherwise, it just means you’re ignoring what they wrote.
There’s also the issues of expecting people to ‘prove a negative’ to accept your claim, which is part of why no one has hard ‘proven’ otherwise despite the associated evidence.
The match history on your, and other players, aram games have sub-50 accounts with less than 50% winrates at ‘master’ level aram, and they’re on teams that aren’t in a party (to break matching)
brawl didn’t use mmr, and had ai bots that could fill for player slots to get quick matches.
Aram took over brawl and has ai bots that can fill for player slots to get quick matches.
While people point at qm matching quirks, those instances tend to rely on heroesprofile – which doesn’t have correct info for mmr – or on ranked portraits – which doesn’t reflect mmr in other modes – and tends to neglect party compositions when it suits the claim/complaint
If the means to make a case stems from ignoring anything else that doesn’t agree with you, the only thing you’re doing is announcing you prefer confirmation bias, at which point it doesn’t matter what others post if it doesn’t agree with what you want. Similarly, the demand from the player set you’re looking to get is more confirmation bias.
Additionally, the usually outcry for modes that have mmr are the ‘feeling’ of having ‘forced 50/50’. If the same methods as your applying lead to the conclusion that the matching is rigged, then that would suggest that another mode would therefore lead to a similar feel – which would be contrary to the success you’re describing.
If aram had mmr, then you’d have smurfs that are getting ‘correctly placed’ in far fewer games than any other mode, don’t experience elo hell, and don’t adhere to ‘forced winrates’ despite a lack of party abuse. So either aram has filtered out many of the bad-eggs that ruin other modes, or it’s matching is that much better than all the other modes, that it’s a wonder why it hasn’t been used to fixed qm/ur/storm at this point…
yea, that’s more ‘not proof’ speculation, but there’s only so many quirks to suggest that confirmation bias is being confirmation bias.
I won’t allow a large margin of random sampling error.
This is not biased.
With enough data, there’s always a solution close to reality.
If the hypothesis fails, then I’m fine with it. I’m not attached to my thoughts.
If I’m asking master players and not others, it’s because there’s a huge disparity of skills between them that will make the experience more obvious.
If someone is doing a survey of pain from 1 to 10 during birth, he doesn’t need to include men. Get the hint?
There is an even easier way to test it. Play a bunch of matches on your main account, the more the better. After every match, look at the profiles of everyone on both teams. Since your main account is currently Diamond, every solo person you are matched with should have a large number of matches, and be close to your rank. If you find yourself getting a large number of new accounts, less than level 20, or Bronze players, then clearly MMR isn’t coming into play.
As someone who actually took statistics, I just physically cringed…
Either way, do what Hoku said.
That will give you the answer you seek.
Hint, the accounts are going to be all over the place because ARAM doesn’t use MMR. 
Yes, ARAM does have MMR.
It’s just a constant tide though.
I am not sure that this methodology would actually test only the existence an MMR system. What you are testing with this methodology would be more something like:
Implicitly, you assume that the system would work as for SL. But it could also be that it is not the case. Currently, SL does not allow 2 leagues apart within a team, but it is in some sense a parameter of the system. They could as well set it to 1000 leagues apart. In this case, a “master” could still be matched with a “bronze” while the average MMR of both teams are the same.
Since I think there is no MMR system for ARAM in view of the data posted in this thread which are more in line with no MMR system (also I don’t really see the purpose of implementing an MMR system for a mode that seems to be a “fun” mode from the start and because of the history of the mode), I would rather perform another experiment:
Here the essential assumption about the potential MMR system is that its main purpose is to average the MMR of the teams. Another assumption is also that such an MMR system does actually its job. Actually there can have some other assumptions (and biases) in this methodology as well: it is really difficult to tell. For instance, if we assume that there is no MMR system, the “randomness” of the teams could as well be implemented as “take the palyers in the order of their queue and distribute them into the two teams”. I think that this implementation would leave the players with the impression of randomness of the teams even if there is actually no randomness at all. So even with this methodology and with this implementation if the players queues in the same order (for example because of their reaction time, because of their respective distance to the server…), they could end up with some form of “fair” games even if there is no MMR system.
Actually, another way to check is to play ARAM, check the profiles (in game not with a third party site) of the players in both teams and looking at their win rates in ARAM (and number of played ARAM games to see if their win rates are repesentative) and look for a game with an obvious discrepency in the mean win rates. I know that win rate is not equivalent to MMR (if the system exists) but it seems to be a good proxy if they are far from the average 50% (and stabilized).
Edit: also if these fails, it does not mean that there is a MMR system. But given a high enough number of games, this should tell “hum… there might be something”.
Checking profiles isn’t telling the whole story. Half of their games could be in a 5 stack with smurfs while the other part is solo queue. It makes the r value look like a cumulonimbus.
Winrates should not matter at all in this game since its not a 1vs1 game.
I agree that win rate is not MMR (again if this system exists). Assuming that an MMR (exists and) tries to equalize the average MMR of both teams, the only thing that matters is to “measure” (measure in " " because we don’t have it, so rather that measure, I should say “have an estimate of”, “have a feeling of”…) the difference in the supposed average MMR between both teams. It actually does not matter how the players achieved their supposed MMR. It could be 100% solo queue or 100% with 4 smurfs teamates or whatever.
“Having a feeling of” is a valid data in my book as long as there is a significant amount of games played.
If I was extreme about it, I’d directly ask these players (sorted with Player MMR) to play cross region on their low level account in solo queue and report their gut feeling on the difficulty of their games. I’d ask specifically the ones with the highest MMR and win rate around 55% to 60%, meaning that they don’t usually play in a 4-5 stack.
喰音 would be the perfect candidate.
I don’t what you want to do with this question but if you want to ask for the difficulty only for the beginning of these good players new account, it will not be enough. You will not be able to distinguish between the two scenarii you want to look at. Here I think you are coming back to the original idea you had in your opening post: following a player through time. In the case of no MMR, a player should find the difficulty of his games to be similar through time. In the case of an MMR, a good player should find his game more difficult; a bad player should find his games easier; and an « average » player, average here with the meaning that the player final MMR is the same as the default MMR, should find the difficulty of his matches to be unchanged. This approach can also work but you need to question the evolution of the difficulty through time and not only at the beginning of the account.
I, myself, personally, prefer the approach I presented before because this approach needs some statically elements; statistics I try to avoid as much as I can because I am not fully aware of the technicalities. I prefer the previous approach because it relies much more on logical reasoning. This is personal taste and knowledge.
The old scenario isn’t disregarded, it’s implied. The focus on the new account is simply to introduce a new step to take part since the main account should already have the data required that include the stabilized MMR.
Like the one player that I’ve pointed out doesn’t need to play 100 more games on his main account. He already have a stabilized MMR, if there’s such a thing. He just needs to play on a lvl 1 account for 30 games and report his experience. If the master player is having 80% WR while it’s 55% on his main, then there’s a chance that MMR exists. Of course, we need a bigger sample size, but it could be the start of it.
Ok I thought also about the main accounts and thought the same as you about their « data » in this approach. I was just unsure about that since you didn’t mention it. All in all, we agree
Spazzo (04/23/2021):
aram has an mmr system, it’s just it’s pretty much ignored
like it’ll make a match before it’ll make a balanced match
but if there’s enough people in queue to make a balanced match, it’ll do that
it lessens very, very quickly
like quicker than qm
And an old reddit post.
Eh I was sure you only wanted to be right! (This is a joke obviously).
Edit: aren’t we the same person with 2 accounts responding to each other to bump this thread?