Or, and stay with me here, when your MMR rises, it puts you with other players that also have that same MMR. WHAT!?!?! I know, it’s a wild concept, but if you’re crazy enough, it JUST MIGHT MAKE SENSE!
Also, back when they had SR decay, you would see players in lower ranks playing in higher ranks, just because their SR had decayed, but not their MMR.
That doesnt make sense - only because of the overall win probability.
Never at any give point will it actively seek out similiar MMRs - the player lobbies is based on random people clicking queue for game.
At some point MMR is treated super accurate, but then they stick the new players in the middle… any MMR is a guess…?
I personally think you want to be in a situation where the game thinks you suck, and it gives you a high MMR person because you are the weighted chain. Then you get the win streaks cause your with people who have the “higher” MMR.
The next day it finds your MMR is better than it thought, so now its like, ok now YOUR the carry and you just cant… and you lose lose lose. (but its based on time of day - which day - which event is running).
I think it would be a grave mistake that the MMR of a dps is mixed in with the MMR of a tank. I think thats why we are having some really bad games because the MMR diff of the tanks are SUPER tilted but it should have split the tanks… (but then comes the idea about teams), if you have a tank duo and the game CANT split the tanks.
I dont envy the problem, but I do believe they SHOULD make more experiments and communicate that. I also think they should have a team channel for folks who do teams. They should average the MMR per role and not just per team. If they do that now (?) then lord help us.
ideally you want to put yourself in a situation where you re able to perform consistently - so no you do not want to be the potato trust me
what i mean is if you are able to outperform your rank time and time again - whether you win or lose - you MMR will go up and you will climb(im only talking abt plat and below bc pbsr + that’s all im familiar with. for now :DD)
yes you might get a bit worse teammates bc you will be the ‘carry factor’ but a couple of losses in the grand scheme of things dont rly matter or hinder your climb
if you constantly ‘prove’ to the matchmaker that you are in the wrong rank it’s gonna start to push you up ‘where you belong’ - this looks like 21+sr/win and less than -20sr/loss(pssst one way you can artificially raise your sr is playing when the playerbase is smaller, morning soloq is worth it srsly)
i think iostuck said that joining a group makes the system treat you like an unknown since it’s harder to determine it in a group(dont quote me on that one cause i cant remember) and idk how true it is(bc of blizz s lack of transparency) but i did find that if im on a bad loss streak(like now…rip my 300sr XDD) joining a group for a match and going back to soloq can send you on a winstreak again
but yeah i agree some clues as to what the mmr rly judges would be nice. but i also can see why theyre reluctant to say anything(cause ppl would try to exploit the system)
bit off topic but has anybody’s matches been kinda weird lately? like i mostly soloq and i recently got a few matches with plats and bronzes on the same team(i checked and yes the plats were on the way down to gold and the bronzes were almost silver but it seems to me the population is dwindling and the ‘acceptable’ mmr diff range broadens…which is. uhhh not great…) i asked if they were grouped and everytime they d claim they soloq(which obvs they could b lying) i realize i dont have any proof to show for it so it’s just like. rambling but i was wondering if anybody had the same exp? i also had a few matches where i was matched with a full bronze team and there was me and another almost gold on the enemy team…what’s up with that lol
Your right that it is fair matchmaking, but it is unfair ranking (really just makes it frustrating) and which really is the number we all care about. The problem is that it uses two numbers one if which is hidden. A true honest system would only use 1 which is SR. I get that they want people to get where they belong faster, but by allowing such a shortcut with ai that will not be a perfect judge you only introduce imperfection.
Mmr would be fine in Quickplay, but not in competitive. In competitive games you already group players within an sr range, so I ask you what do you need MMR for and why is important that it looks at your past games, doesn’t your skill rating represent your skill and wins/losses already? Obviously everyone who is SR 2000 isnt exactly of the same skill level, and that unknown variable is what mmr tries to calculate use.
Let’s assume all 12 players were 2000sr what mmr will try to so is even the team out balancing them out. The problem with that is that the hidden variables which mmr attempts to manipulate are what competition is it’s the unknowns that are used to seperate players of different skill. Manipulation of it promotes rank stagnation because the closer you get to achieving perfect 50/50 game chance every game the harder it becomes to seperate the players of different skill.
Why do you need MMR?
Well simply said, MMR is everything.
SR is just a simplified version which can be split into multiple rankings so we can like read it.
Make SR do everything and all matchmaking is gone.
You could do it around an SR range, though that wouldn’t make matches difficult if you were to win too much to compensate.
You’d also have a static number that wouldn’t change depending on statistics on so on, which would most likely be 25 SR +/-.
Matches are supposed to become harder if you win too much or become higher, that’s how a ladder works. If you start winning wayy too much then it could be balanced out with players within the same ranking, showing that the rank you’re in isn’t as easy. This way one team won’t constantly lose everything.
And well the higher in SR you go the higher it becomes.
Honestly i don’t see what there is to gain from SR, it would only focus on winrate.
And winrate doesn’t always give fair matches.
Competition IS unfair, the unfairness is what separates the winners from the losers. Competition depends on the discrepancy to exist. Naturally in a competitive sport this unfairness is unknown and only those within a general skill level such as the NBA teams face each other but the are pitted against each other with no mmr manipulation, all that is known is they are within a range which is similar to what pairing by your SR Rank already does.
Tell me would you enjoy competitive coinfliping? How accurate of representation of yourself is captured in competitive coinflips with 50/50 matches? That’s basically what MMR tries to do is get it so even that skill seperation is less effective.
I’m not talking about the SR boosts at the end of the game (although that is also redundant in any true competitive sport) I’m talking about what a previous poster quoted in which mmr is used to balance teams on top of what sr rank grouping already handled.
You do not understand what you quoted.
Read the whole post.
Why the heck is a post so awfully long to explain something as simple as a ranking system should be. Perhaps it’s written to be super layman friendly but I prefer to get directly to the point so perhaps you can do me a favor in telling me the point.
For example I will give you the ranking system of mostly every competition thats ever existed. If you win you move up if you lose you move down You get paired with people who have moved up as much as you have moved up, This rating is generally what your rank would be considered. traditionally it does not matter by how much you win a basketball game for example you only progress or you don’t.
See, an honest ranking system is simple like that.
Ask yourself…why would it do this??? Why would it not just put all people of the same MMR together??? You know, like…a ranking system.
What purpose would it server to put a 10, 9, 8, 7, 6, 5 on the same team when you could just put all 6’s?
Again…why would they do this??? And why would time of day etc. matter?
It’s long because people want to know details. If you read the summary you will get most of it.
The ranking system you describe leads to slower convergence to correct rating than what Blizzard has implemented.
Cuthbert, I recently started a YouTube channel (about time I know), and I wanted to know, if you would like to do an interview regarding Handicapping.
I have no idea how else to contact you, expect thru a reply here.
If you’re interested let me know. We’ll exchange info later.
Yup that is correct but based on stats they cannot decide between lucky ults etc. They would need an AI to investigate all key presses and positions to judge it unbiased. Just remove all handicapping/fair matchmaking/algorithms and we are fine.
Reading the summary did the opposite as it only reinforces my point.
Explain how? As far as I know its slower the way blizzard does it. Halo 2 for example had 50 levels and did it the simple way, from my experience playing that game it was a lot better than what blizzard does and far more accurate. Blizzards system will have fluctuate within 500 SR the fluctuations i experience halo were only ever 1 level up or down which is the equivalent of 200 SR. Personally it’s these fluctuations that I think bogs it down so much because I reached my peak rather quickly whereas overwatch takes random dives down.
Blizzard won’t release the details however and that is the most telling. Why wouldn’t they? Being a videogame programmer myself (Java mostly) its all suspect to me.
It makes almost no sense to attempt to improve and what would naturally occur had they simply used the SR Ranks to group similar players and kept a standard win/loss on SR. While I understand what Blizzard was trying to do, unless you can guarantee perfection whatever you implement will bring errors and unbalance with it even simply Grouping by similar rating is not the perfect system, the very nature of competition works best the to seperate skill without human interference.
For example if you and I were to fight no rules we would go at it with all we had and their would be no doubt who was the winner, but if as soon as you add or modify anything you potentially handicap or boost a player. Of course no rules could also mean getting paired to fight an infant would be fair too so that’s what the whole rank system seperates in other to avoid extreme situation. However if you attempt to balance it by adding an adult on the babies team then I could potentially lose and now maybe next round the baby moves up into a rank it doesn’t belong in and I go down into a rank I don’t belong which is why it you sort out skill better the less you mess with it. The less you mess with it means a higher potential on unfair games initially BUT the faster seperation of skill and the maximum accuracy potential that will increase the more games you play
I will say one thing however, it could be that Blizzards systems would have worked a lot better had not it been for alt accounts which largely were not present in the Halo 2 days. However that would only point to another issue that Blizzard need find a fix for (understanding it was neither their fault or error in this regard but that’s the way the world is and the problem is now in their hands)
Blizzard never responds to player messages on this forum
That doesn’t make anything in your thread facts, just speculation.
It’s speculation that the sun will rise tomorrow. It really is. But I’m going to go ahead and bet on it.
Just like the rigging in this game. We have devs on record talking about MMR. We have on record that it exists, and that the MM uses MMR to try and rigg 50/50 matches. We have the patents. And we have the SBMM precedents for CoD/ATVI products. We have the public outrage on record, as well.
The very existence of MMR implies:
There are parts we don’t need to speculate on, because we have math. We know which model structure is sufficient. We know an SR-only ladder is all you need to rank by SR. It’s all you need for provably unrigged matches.
Stop with the lies, you do not work for Blizzard/Activision, stop spreading false information. Just stop, you have no idea how it works, and neither do I.
There’s posts dev bloggs and interviews detailing some of the things the MM is doing. 50/50 is on record, Scewed matches is on record, prioritizing queue time is on record. And then we have our own experience and logic to work out the implications of this.
50/50 on record, give me the link because I don’t believe you, and everything based on player experience is speculation, what you experience isn’t the same as what I do, and this goes for every single person playing this game. And even if 50/50 is real, it’s still not holding people back, everyone is getting to their correct rank, slowly granted, but nothing is instant, and you would be a fool to believe it is.
For those of you who think SR based matchmaking only would solve problems, don’t fool yourselves, it’s still going to be 50/50, the difference is, MMR will boost gains/losses based on skill, where as SR only would be based on a win rate, not a skill rate.
Quote from Scott Mercer:
We’re always trying to put you in fair matches that we think you have a 50% chance to win, and do it quickly so you’re playing Overwatch instead of staring at a queue timer. We never intentionally seek to put you in an unfair one.
htt ps://mystgraphics.com/overwatchforumarchive/competitive-season4followup.html#post-13
That’s the confirmation. Rigged 50/50 matchmaking right from the horses mouth 
Took me 30 seconds to google it, you can too before you assume people are lying.