Yup, I got myself tangled about on that one. Core problem was that I just accepted that SR and MMR can diverge without going back to read the actual info.
There’s even a theory (based on Blizzard statements) that for Competitive, there isn’t an MMR any more. There is only SR.
I think some of you are mistaken how matchmaker and mmr is actualy working.
Based on my experience i belive that MMR and sr are actualy very close to each other and mmr is updated after every game . Matchmaker is creating games based on few things, but most importantly based on MMR. But you can clearly see how teams have balanced SR usualy at start of every comp game. Why is that? Because mmr and sr are actualy very close to each other. And i think that matchmaker doesnt really see far than +200 sr. You can see it in initial placement on new accounts. You will start in gold and even if you perform extremely well and win your next game would not be higher than 200 sr.
So, matchmaker is trying to create balanced games but he simply doesnt know if person is plat smurf or gm smurf in bronze elo. If top 500 player is in bronze game, his mmr after win will not be suddendly +1000. Its not like that at all. I could clearly see that when i was in bronze on alt as high plat player. Even when i was performing very well and hard carrying, i wasnt suddendly placed with people from silver next game. I went up like +50 in sr and my mmr went up just by that number or slightly higher number too probably. I surely dont have exact data but its even possible that mmr is basicly same as your sr and is different only in case of leaving games,placement and decaying.
This is just my theory, i will maybe create special topic for that, but if thats how matchmaker is working, its reason why players are so frustrated with smurfs and thinks matchmaker is rigged. I say its not not accurate because system cant see difference if smurf in bronze is plat or grandmaster. And if you have luck in being placed with plat smurf while enemy team have gm smurf you can do only little to win.
Its not rigged , it just unable to create balanced games with so many alts anymore and games feel like its luck based for many people, you will have 10 nice games where you climb and next day 10 games losing streak. Dont get me wrong, you can climb , but for someone who is getting better slowly and is not smurf on alt already, is climbing very painfull experience now.
I definitely feel shift in how people behave too on top of that. They dont care that much, because many people are on alts. People too dont care about team comp as they were few seasons back. I remember people being tilted just because we had 3 dps and one tank. Or simple because we had widow. Now you can often see 4 dps no tank and nobody will react, even when they know its most likely loss, but nobody cares anymore. Or only few people truly care.
Found much the same myself honestly.
I don’t play my forum account much anymore. It is a fairly old account and significantly lower than my alts. Yet I get nervous when I log on and find myself starting to care even though the matches are easier.
With my alts I kind of disassociate with them and just don’t really care about team comp etc, I just instalock and the odd thing is I actually went up another 2 ranks with them. It was kind of a weird way of learning that I needed to just let it go and stop worrying about flexing too much which was holding me back.
5 DPS? No worries, I’ll be the 6th with my alt because I know I will pull my weight. I’ve won a lot of matches like this whereas I’d be stressed about losing SR on my main.
What happened with me was I started getting players that were below level 30. Basically players with new accounts. I had over 80 kills in a single game because of it.
Trash game is trash.
No, because all this does is create one match of abnormal difficulty for the rank in question. The chance to win for both teams is 50% (not taking into account which outperformer’s hero choice has higher carry potential). Over a larger number of such games, neither smurf will climb and will stay at the same SR, continuing to distort difficulty for the games hes playing in.
This is how it should be. The outperformer ROFL stomps the enemy team and proceeds to climb out of the rank he is “doing damage to” for every game hes is playing there. Of course, hes giving his team mates a free win, but this should even out with the free losses they get from outperformers being on the enemy team. Outcomes determined by smurfs suck. But thats a different problem.
As you said, thats unlikely to happen due to availability. Also, it would essentially be an equivalent case as the OP is describing (MMR mirrors to force 50% win chance).
Again, same problem because its forcing a 50% win chance by matching MMR outliers with mirrors.
Solution 1) is the only correct one, because
a) it allows overperforming players to climb, which is what the deserve, and a ranking system MUST make sure these players move to a higher rank because its otherwise producing UNFAIR and UNFUN matches.
b) it prevents match experience at a given rank to vary DRASTICALLY because it would create a high MMR battle field where many/most of the players are just coleteral damage => UNFAIR AND UNFUN
Of course, on the other hand, solution 1) would produce ONE-SIDED matches because outperformers are not mirrored, but allowing these players to “pass through for free” is healthier for the rank in the mid/long run.
Speaking about smurfs in particular: I was using the termn “outperformers” to avoid the termn “smurf”. I think a smurf would be a player who’s main account is multiple ranks higher than where he is currently playing. The reason why these players are not having trouble climbing with the current system is that their skill is much higher than what i call the “outperformer”, who is merely performing at several hundred SR higher than his current rank. Also, the system will have a hard time to actually find mirrors for a real smurf in a reasonable amount of time.
I went down a thought exercise based on flawed assumptions in the OP’s statements. SR and MMR do not drift apart except in the cases of DC penalties and Decay.
Alts and smurfs are different. Alts just haven’t gotten to where they belong yet. Smurfs have had their SR/MMR intentionally manipulated, and damage everything about the matchmaker.
Fortunately, in Silver+ the Alt/Smurf problem is small enough to only impact people occasionally, and it’s random whether it’s a positive or negative impact.
Smurfs unfortunately have a very outsized impact on low bronze (500-800SR) where the thrower trolls go to destroy games endlessly. The population is so small there that they have a significant impact.
Repeating: SR/MMR do not drift apart even for throwers. The matchmaker does not do anything but attempt to match 12 people with the same SR/MMR into a game.
I’ll be blunt…I have watched over 300+ hours of progamers, none of them popped off to anywhere near the level I did in any of their games in GM or top 500…ever…I think you need to go watch some bronze to top 500 videos to see what mismatched MMR really looks like. I just climbed another 120 or so and I started a fight 2v6 killed two in the time my teammate took to die and then killed the rest solo, solo pushed the cart, then the enemy respawns hit and I killed 3-4 more before my team got back.
That’s not something that is nearly impossible in plat and above, and can pretty much only done by GM/top 500. winning a fight that is started 2 down is already pretty uncommon, winning fights 4 down before the fight starts and then continuing and cleaning up their team two times in a row is only possible with horrendous mismatches.
I am not complaining about the climbing…I’ve already gone up roughly 400 in just a couple afternoons, but I question whether this is really the best way to handle smurfs that need to be pushed to a higher MMR/rating. If the MMR is already so high why can’t they just put me up against plats by now. you’d think with all the wins and stomping it would be time again, but no instead, it looks like I have 3-4 more hours of play to get to plat.
Luckily I am finally out of silver/low gold so I’ve been pulled back from popping off at levels above what we see in pro games down to a more reasonable oh I’m just carrying 2-3 people’s worth.
it’s not even good practice it’s just burning approximately 6? 9 hours of my time? I don’t really keep track. I do other stuff at the same time. But as a method of keeping players engaged and playing more I guess it’s ok, but doesn’t seem like a good or fair method to me. At least if I were up against plat players by now I could start trying to learn the characters I plan to play on this account. however there is nothing to learn playing at this low rank except that you need to be supplying primary dps regardless of role because no one is doing any damage. I really am curious whether ana players can maintain 80-90% win rate in low ranks because it seems like it would be extraordinarily hard. But maybe they just don’t heal and go full on kill the enemy team. I feel like they will win the games in which they have a second main healer, but…
sometimes I think you could be right, but I only see hanzos killing 4 every fight and hitting extremely hard shots that a gold would never hit when I’m on an alt account and outperforming. I don’t think that’s coincidence I think they match a high MMR player to match me. If I just am sure to underperform and keep my win rate even then those players don’t ever appear
It sort of does. In the case of iddqd (allegedly) they stated that his MMR on his bronze account matched that of his main after 15 games.
This is possible, but it’s also possible that you will be placed in higher ranked matches where you’re the lowest SR player. Toxicity occurs regardless of MMR - it’s an uncontrollable variable.
They already do. Once the system determined I was a smurf I got much more SR per win than I lost for a loss. Sure, it takes time, but this is to prevent errors in MMR calibration. If you can perform consistently for a large number of games then it will determine you deserve a higher rank.
That’s nowhere near a large enough sample size to justify gaining that much SR.
I believe they already do introduce test games every now and then. I’ve seen it myself plenty of times, but I’d need to check more sources for this.
Damn! watch out OWL !
No it’s not that I’m good. Owl players pop off similarly in low rank games if they try.
Over 500 point difference is outclassed and 1000 is being worth 5 of the enemy team in most games. Anyway i guess they just don’t want boosting to be easy. I was forgetting people do that.
Why should they be different? Both types (in my case a) outperformers and b) smurfs) have an internal MMR (or any kind of performance metric) which is higher than their current SR. Both should be treated the same way by the matchmaker. In the sense of the OP, thats “mirroring” their MMR to create a 50% matchup.
I know the debates weve had on the forums in the past and the claims that MMR stays very close to SR. However, im not so dogmatic about that. There is certain evidence that this is not always the case. For example, i heard that smurfs get huge SR adjustments for games, and that these adjustments are incremental, meaning, if they go on a streak, the adjustment increases from game to game. This is one thing that would support the thesis that there is some performance-based value (be it MMR or another value) which is decoupled from their SR, because that value acts as a “memory” for how well they did in recent games (SR adjustment calculation is not solely based on a per-game basis).
If we do have cases where MMR (or whatever value) is not close to SR, and if matchmaking takes MMR into calculation, then we can have cases where MMR mirroring is in effect, as the OP is reporting.
According to Blizzard (via this very thoroughly documented article) MMR might not even exist in Competitive any more. If it does exist, it matches SR in all cases other than disconnection penalties or decay.
That’s PBSR, and it’s not an account flag, it’s an adjustment based on the calculated win chance, and an analysis of the statistics you put forth in just that game. (see article)
According to the article, PBSR is intended to be relatively small, and again, based on statistical performance of a single match, compared to the average statistics for the hero you played of the people in a similar SR/MMR.
I can’t actually find where he says MMR may not be used anymore. I’d appreciate it if you can link that.
Nope, that’s MMR itself. It uses your MMR to calculate your win chance, and your SR will change more/less quickly dependent on how different they are. Also, all my accounts are diamond+ so there’s no PBSR involved for me.
Jeff says here that smurfs are moved out of ranks quickly as MMRs on them will match the same as that of the main account of the smurf. It’s kind of an account flag in a way.
That’s because of all of the following at the same time:
1 - The new account “uncertainty”. New accounts have much larger SR/MMR gains and losses after completing their placements. You’ll see them with base gains or losses in excess of 50+ after a single match (not even factoring in “PBSR”) and those base gains/losses taper down as that account plays more games, and then by around game 50 or so, the base gains/losses taper to around 25.
2 - Performance-enhanced MMR gains on a win (what people like to call PBSR but I see it as a misnomer) since that explanation was made at a time when it was still active for all ranks (April 2017)
3- Winstreak bonuses (which manifested itself in seeing the maximum 150 SR gain on a win), which are now nerfed
Dafran recently did an unranked-to-GM stream I think. I didn’t check his win/loss record for that account, but however fast he did it is about how fast it can be done in these current conditions.
None of my accounts are new. The one that’s in masters now was created in season 5, and regularly hovered all around diamond, and even fell into plat the previous couple seasons. This season, I was gaining around 26+ SR per win and losing less than 20 per loss.
MMR is most probably based on your performance relative to your rank and the game you’re in. Sure, as your mmr rises you will gain more SR, but I feel that PBSR is a more short term variation. The system will find matches based on your MMR and adjust your SR gains accordingly. If you perform better, your MMR rises and you gain more SR. This takes at least 15 games as shown in the previous post. If this is the same as PBSR, then I don’t see why the devs would make a distinction between the two. I think PBSR gives you a small amount of additional SR extra dependent on your performance in each game, whereas MMR gains looks at your entire season history. I could be wrong.
Scott Mercer says here that they tried to make it only kick in now when the matchmaker was certain that there was a discrepancy between current SR and MMR, i.e. a smurf. The maximum 150SR gain is very unlikely to come through nowadays unless you’re a diamond+ on a bronze account.
MFW people still think matchmaking is rigged.
https:/ /i.imgur.com/ki79mAg.png
Like … how much more accurate does it need to be xD ?
Sure you get aids teams/games from time to time but you will end up where you belong. Doesnt matter on which account.
yeah I’m going…going to be in plat by mid-next week, and I can already feel that my ashe can barely win at 2300 and is like 2500 max, same with my widow. sniping is hard to carry with compared to mccree or soldier where I feel the games are super easy until around 2800…
It’s just I had to play way too many games in silver and low gold…even with 70% win rate…and it’s too hard to maintain 90% unless I always fill because as your MMR goes up you get harder games, where your teammates are heavy~~
I swear that’s what happens. anyway not a big deal. yesterday I had 100% win rate over 7 games, but that’s cause I’m not only playing at around a diamond level but also filling every team comp’s needs…
I have a feeling if 2/2/2 lock really comes that suddenly it won’t be so hard to carry low mmr players because at least you will have a team comp with 2 underperforming players instead of getting 2 underperforming players and then having 3 of your teammates swap to dps as if that’s going to help (to be honest though I think optimal might be switching one teammate to dps and leaving one realiable player on both heal and tank so you have all bases covered). However, that’s not what happens. they switch to 4-5 dps and are practically trying to lose. even if my orisa would normally carry 70-80% of the time they just see the dps dying over and over and believe the game is unwinnable without more dps
Not sure if my situation is the same as yours. But I have been pulling good Zenyatta plays since yesterday (for the very least, for a measly Silver, my play was really good). I got important backline picks, timed my transcendence and discord much better, kept my teammates alive, and almost always had a card at game end dedicated to myself (either gold healing, gold damage, gold elims, trans healing, defense assist, or offense assist) and I’m also starting to get more and more POTG as Zen! Which I consider huge in a rank where DPS heroes constitute at least 3 out of 6 members of the team.
Anyway, my SR is currently sitting at 1800+. Weirdly enough, one of my games last night, I was matched with 4 bronze + 1 silver players with average of 1200+ SR. Enemy team also had the same SR average but they have 4 silver and 2 bronze players. Most of them had their career profile private but one of them who was on my team, his SR was 1084. Not that I’m complaining because I was taking their snipers (Widow, Hanzo, Ashe) out on a walk. We did end up losing the game because one teammate left, followed by the other two, leaving myself and the other two teammates to fight 6 people.
All in line with I have discovered myself, especially the “mmr reset” within a few days of not touching the game, playing less also makes the game a lot mroe unsure. Discovered that when I only play 1-2 games every two days, I get an 60%+ win rate on average regardless of SR.
a coin has a 50% win chance, throwing it 50 times doesn’t mean you get exactly 50% wins
basic probability
how does it work!?!