For the discussion of “mmr changes in line with sr, the two are interchangable, just sr is a simplified version of mmr” there are two factors we need to ignore:
decay, as this is a unique exception by design
leaving/disconnects, as this works in line with decay
The discussion has to be made assuming the player has already matching mmr/sr, otherwise it cannot even begin. Those two examples are outside that starting point so are moot.
You are claiming, that on a win or a loss, from starting matching mmr/sr that the player will continue to match mmr/sr. With the limitation of placements, for no discernable or cited reason, other than that your model does not hold during that time.
I feel this is going to devolve into semantics, which will be completely silly.
That only happens because the sample size tends to be one game with that team of random players. If the same 12 players would play a bunch of games in many cases they would go back and forth match wise. I use to see that all the time before my friends list more or less emptied of players. Being I use to be able to 6 stack in comp or QP and you would get the same teams over and over some nights. You roll them and think they are trash, they smash you like they are all gods the next time around on a different map or play mode.
Like my stack was trash at KOTH but monsters at payload. Yet the match maker doesn’t take that in account.
as for MMR vs SR, the devs said on stage at an event that MMR moves slowly and SR moves quickly so unless we hear other wise that should be viewed as the default mode.
Individual performance varies wildly between matches and heroes. I have games on mccree where I headshot everything and get accused of smurfing, and other games where I get accused of throwing because I can’t hit anything. Performance also varies by hero - maybe I’m a diamond Zen but a gold Lucio.
Also, Overwatch is a very steamrolly game (especially at lower ranks). If the enemy wins the first fight, often your team will keep trickling into 1v6s and you’ll lose the rest of the game. Players also often get tilted after a bad first fight, which puts that team at a disadvantage.
I haven’t seen them specifically saying this, can you provide a link or enough info I can hunt this down? Had a look but couldn’t find it.
Instead what I find are two comments, that SR always moves while MMR can stay the same, and that the more you play the more certain MMR becomes and the less you play or if you take a long break it becomes less certain. Neither of these preclude that MMR can move faster than SR at times.
I am not saying it always moves faster. Only that it has the ability to, and the times that I claim MMR moves faster are when you have a complete stomp match streak. Either 4 or 5 stomp wins or loses, wins where your role feels super easy to fill and you are hardly dying - or loses where the dreaded “3 kills is gold” happens.
I’ve looked for the dev talk before and there is so much other chaff when it come to google terms like Overwatch, Dev, MMR that it’s just lost in the sea on info. It was a stage talk of the dev on stage, redish background, it’s not blizzcon 2017 where they talk about the back ground of Overwatch.
There was a lot of other people up there including the main math nerd for the whole system. As Jeff joke that the math guy was the only person who really understood the system.
Keep in mind most of the big MMR/SR talk was well over a year ago when Blizzard dropped the bomb on everyone that we were being judged by an automatic deep learning system that pulled data from millions of games. Then compared your current out put vs tons of other players on that map, on attack and so on. It’s not based off just beating the players in your current game, not really.
some how over the last year or so as new players roll into the game the whole MMR/SR issue comes back up for some reason. The way to really dig into is to look back over a year when all the dev posts where made along with all the youtube vids about it.
Now? At that point most of us have a least a baseline idea of how it works but not the raw numbers/stats that are more important. That’s good being it stops stat farming.
Almost all the rabble rabble tinfoil hat stuff for the match maker, the idea of fixed games, forced lose etc are all lower SR players who think they are performing at +1 or +2 higher but in most case just aren’t doing so.
As show with VOD reviews of lower SR play and how every time they make lower SR plays and errors. Every… single… time.
Actually there are those who have been trying to determine what it was that felt off about the matchmaker in competitive since it first came out. I’ve been playing since season 1, love the game, but can see massive faults in matchmaking that are there to prolong game time and loot box purchases (there are patents from Blizzard/Activision for exactly this purpose). I’m just saying it isn’t new or cyclical, it’s been constant.
The VODS are only done on the games that are clearly as you described, so selective evidence will always push one way. Here is a link to S23 providing an example of a GM being super handicapped in Bronze to the point they lost:
No, it’s more like bronze players playing like Bronze. Yet a VOD review of the game would show the GM player clearly not playing like a bronze and everyone else doing so.
You can’t win every game. A lot of GM level players get new accounts, place in say high plat and then slowly work back up to GM. They lose games along the way, it just happens.
yet over time they reach their so called water level again.
The problem is that most players can’t be objective about their game play and look to blame others and the match maker. Yeah it’s not perfect but if one can’t move up after playing 100s of games in a season, then it’s the player.
This is the whole point, I’m not sure if you understood it or not. As you watch the 2 hour video, his games get progressively misbalanced where he has to carry the team. By 55 minutes in the GM is in a game doing 17k damage, sitting on 4 golds by what you can safely assume to be wide margins, but only just squeaks the win in overtime.
The GM is playing like a much higher player than Bronze, yet is in a game that is hard to win.
After that the matches get even harder, until he loses.
He isn’t getting harder games because the players all around him are improving, the game is trying to balance his now busted MMR for his SR to a 50/50 game. He can still push it a bit because he is actually much better than both the MMR and SR limitations.
However, imagine a Bronze who just had a lucky run of games being put into one of those matches instead of the GM. It would have been an unbelievable stomping defeat. MMR isn’t working correctly for balancing games if this example exists.
There is no discussion from me on “I can’t climb it’s the matchmakers fault” and I have never claimed this. At least I don’t think so, since I did climb Silver to Plat back in season 2 when I got my head around FPS after not playing them since UT2k. All I claim is the matchmaker is the cause of yoyo streaks because of how it is handling MMR.
It balances busted MMR by way of the Performance SR system, not so much match making. Gaining more for wins and losing less for losses. Looking to push a high stat level player up and out of lower SR with a lower win rate than a poor stat sort of player.
One of the more popular tinfoil hat theories out there is the match maker gives you “worse team mates” if you win a few games to force you to lose or make you carry.
I assume it’s based off what one might see in very high SR streams where a player like say Surefour and the matchmaker can’t find 11 “hims” to balance out a game. So yeah, he might get a masters in his game being that’s the best the match maker can even do.
that’s not an issue in low SR ranges, at all. Being the player population down there is so much higher. The reason the player keeps cranking out high stats is they are a GM in a way, way too low of a game for them. The reason his team are not that great, is they are lower SR players in the first place. The match maker wants that GM up and out of the lower ranges and odds are is flooding him with MMR/SR gains with each win to do so.
that’s how they balance, that’s why we have performance SR in lower SR ranges. Even if he is getting a few more underdog sort of teams they are not going to be night and day different than another other random bronze/silver/gold player. Why? They still are silver or whatever SR too. There really aren’t “good gold players” and then “bad gold players” for match maker to find. It’s just gold players, who play like every other gold. As in why they are gold in the first place.
Now you do see a bit of crazy team levels in QP. I’ll be the only 3k+ player on team of golds vs a masters, a bronze and a team of golds. That’s a different issue being the match maker can go huge with it’s range in QP. For comp, there isn’t that big of a range of players to work for this theory of “Oh now I get the bad players being I won too many games” or whatever I see getting kicked around.
It’s seems strange to even suggest it when in most cases the player saying it is maybe SR 2700 and so is every single other player in the game.
Yes, they do.
Go digging into all the dev post and break downs of said dev posts and so on. I have, it’s not as complicated as people seem to make it outside of the raw math they hide from a person. All wins are not seen as equal, MMR tends to move slowly, SR moves quickly. The system looks to pull the two numbers closer together. Wins mater more than anything else.
Performance SR takes into account how well a person is playing vs every other player at your SR, on that map etc and if you’re way better than your current SR then you get a performance boost. It’s to help get that player get up and out of the lower SR ranges. That’s why the devs don’t seem to care about alt accounts being they have said the accounts get placed and moved up rather quickly. That’s a different issue from smurfs who throw games to derank, breaking the system.
The tinfoil hat stuff where players say match making is forcing losses, giving worse teams (even when SR is the same for everyone) or that due to early low MMR/SR a person is hard locked at the lower SR. As if some random gold is player at a diamond level but the system will not be allowed them to get out of gold.
yet all of that has been shown to be false just from VOD reviews and also when players give up accounts to friends who are better players. Then of course the better player rockets out of the lower SR range like it was nothing.
It’s not a perfect system, most of all at the highest sub 1% of the player base level that is GM, where it just runs out of players for good match making. Yet it’s also not out to get people the many see to pitch it.
There is contradictory stuff here, so let’s break it down assuming what you are saying is correct:
This means the two do not match. So for any given SR there is a variance for mmr. This would mean there are “good” and “bad” players for each sr.
If your first point is true then this isn’t tinfoil but required to be true.
You don’t see vods of bronze to gm climbs. This is a selective, aka biased, process.
Make your argument without talking about “climbing” and “vod reviews” and I might take it seriously. All you’re doing still is showing a confused platform shown to you by the muddled arguments against handicapping by mmr.
I’m sorry, but “a video I swear I saw, honest” is not admissible evidence, especially on this forum.
I’d love to see a video with the “math guy”, so much so that I spent a while looking for it last night. Do you remember his name or title, or anything else to narrow it down?
Let’s not overcomplicate things. MMR is probably implemented like it is in every game since halo. For SR inactivity leads to decay. For MMR, inactivity will adds uncertainty - it is allowed to change by more following inactivity. Otherwise they are relatively similar. Well, barring the possibility of losing MMR from a win, which they may or may not allow.
The reason top 1% players hate MMR is that the better they get, the worse teammates they have to play with. This is easy to experience in arcade/QP. I imagine it’s not fun.
Mmr is a hard stat that only moves if you improve (based on elims, damage per 10 min, how well you use heros kit etc)
Sr will always float above or under mmr. If your mmr is consistet and never changes (example mmr is 4000), your sr will alwaya be more or less arund 3800 - 4200(example).
There is a night and day difference between match making being “close enough” to get a game going and a deliberate effort by an AI system to single out a given player for a loss. Then stacking up their team with losers. 2233SR vs 2441SR is close enough match making wise to be “even” from the point of view of the match maker while also being off enough in reward the lower player a bit more if they win and perform well.
The real problem is the arrogant view point of some gamer in silver/gold/whatever to look at fellow golds etc and think “Oh match making deliberately gave me the bad team!!”. As if some how the other team of golds (as in below average) are some how this merc team at 2100SR but the player’s team of 2100sr players are terrible, deliberately make that player lose and stay 2100sr. Errrrrrrrrrrrrrr ok.
It’s was a full on dev panel on stage with folding chairs with Jeff, the lore guy (don’t rem his name) and bunch of other people. It was clearly a large scale event but the only one I keep finding is Blizzcon 2017 and them talking about the history of the game.
but as for the MMR moves slow, SR moves quickly. I thought was just common knowledge. Every break down of the MMR/SR system tends to mention it and Jeff has said the classic line of “How the system looks to pull those numbers closer together” before in other dev talks if I rem right.
The MMR is what the system thinks is the player’s true rank, it’s why it’s tough to hold a higher SR unless you just keep winning.
You’re getting pulled backwards towards your MMR, being it’s not moved up as quickly.
how much this impacts the game for average player is of course blown out of perspective. As they are not playing +2 tiers higher as many seem to think in the first place.
In normal circumstances, though, I don’t find clear evidence one way or the other. I used to say that SR always moved faster than MMR, but the data that supported that hypothesis can’t be reproduced in recent seasons.
Overall, though, I don’t have a clear way to measure MMR movement speed, and I have not seen any clear developer statements.
and I’m sure that’s deliberate…
We all want to know the exact details of the system but I have a horrible feeling if we did everyone would turn stat farmer. Sort of how schools can get a bad rep for “Teaching to the test”.
It’s still a little strange to see mmr/sr issues still popping up as the whole rabble rabble over it was from well over a year ago. There just isn’t anything new and ground breaking going on. The last big bomb shell was like a year+ ago about how we were being judged off the huge data base of players on each map, in each role vs the common thinking of it just being the current game and beating those players.
other then that? Nothing for a year and half or so. Everyone sort of moved on to focus on hero balance vs the system.