Why was this called “Part 1” if they aren’t going to say anything else?
I do mean the former. I’m not saying it can’t happen or won’t happen. In fact I think it will happen to a significant percentage of the player base but not systemically. What I’m saying as they’ve described how the system works (today), there is friction in place to get back to where you were (sweeping de-ranks while keeping MMR as is). Because you experienced it doesn’t mean it is a manufactured byproduct of the design of the system. In fact I’m certain there’s a contingent of players that claimed they’ve struggled to get back to their rank despite having the same personal skill level (I know I’ve seen that anecdotally on the forums).
But if they have stated that it is how the system works, do you have an idea of where/how I might find that language?
The strong language I’ve seen though is that MM uses MMR and has no concept of Rank, and thus climbing back to where you were at best is a phenomena of natural distribution across a bell curve given enough games whereby your plateau is a by product of not your historical Rank but a current relative skill of the population, which is erratic over the past few months.
I think the mistake you are making is assuming that because matchmaking is not linked SR, that MMR is linked not linked to SR gain and loss.
If you are a higher MMR than your current corresponding rank, you gain more SR per win and less SR per loss, which is why you can climb even with a negative win rate.
I wouldn’t put much stock in people claiming their skill is the same. I’d rather see them state their win rates.
I would suggest reading this blog post in full: Initializing systems! Updating Competitive play for Overwatch 2 - News - Overwatch
But particularly the sections on skill tier decisions, skill decay and placements.
Here they basically say SR is a way to keep you feeling like you are progressing.
I’m assuming last 20 losses or last 7 wins are used to determine rank changes/progression. Those matches were setup by the MM based on MMR not rank. Agreed? How it is computing the SR change is anybody’s guess?
Is that the reason though?
Isn’t also likely based on information shared that given a 20 game snapshot that had a 35% total win rate that some of those loses were so lopsided (MMR) that they almost no negative impact? This was actually observable when SR changes were displayed after every game. I still have not see anywhere that it has anything to do with “your current corresponding rank”.
Thanks for sharing that link. I had read that but its been months. It’s just vague enough on detail though. The only thing clear there is they are making Rank progression abstract as opposed to game by game measurements and supposedly for our quality of life improvement. The actual result though is that it removes any ability to measure the impact of a given game. I think it’s a big step backwards.
EDIT - I realized I didn’t give much of an argument as to the vagueness. Take this from the DECAY portion of the article.
So it clearly discusses the individual players skill performance. It also clearly discusses how it will use & manipulate MMR. It even suggests MMR updates “as they play matches”. But the kicker is “quickly get back to their appropriate skill tier & division as they continue to play”. You could read that as you were mid Diamond 2 years ago and welcome back and this system will get you “back” to Diamond. But logically, it should have omitted the word “back” for clarity and “appropriate” is a function of future performance not past.
This is how it should have been, no need for all these sub ranks.
Agreed but how it’s computing SR change is not anyone’s guess.
They have already stated before SR “chases” MMR: https://mystgraphics.com/overwatchforumarchive/competitive-SRdecaydoesnotaffectmatchmaking.html#post-16
This is an archived forum post from the previous guard. You can look at MM topics there.
They also mention MMR can go down even if you win but SR cannot. It’s very clear that SR is for engagement and MMR is the “true” number.
Quoting this directly from Blizzard too. This is referencing the current season.
So matches should be fair and yet you should still climb after a decay.
Tell me how this is possible without your SR actively chasing your MMR with greater gains for wins until your SR corresponds to your MMR?
Yeah, it clearly worked that way when SR was displayed every match which I believe existed for several years including August 2017. What I really meant to say by my anybody’s guess is that because they look at either 20 games or 7 games until you see a change in Rank, you no longer have any objective information on what a single game gives you. For all we know now, they are adding additional logic “to make things feel better”.
On the most basic extreme, it could be they are doing exactly the same thing that they were doing in August 2017 under the covers every match, but only showing the new Rank once every X wins or Y losses. Who knows really?
There’s no question that your performance will impact your Rank. That was never questioned. The problem I have is “until SR corresponds to your MMR”. It’s that assertion that is bogus. Now, if you say gains until you have a less than 50% win rate than those with a close but slightly higher MMR, then I would agree. In which case, the side effect will be that you’ll plateau and struggle to rank up further and that defines ones skill level (not Rank which tracks progression). Per your 2nd article (thanks btw)…
I don’t see how “Ranks representing their true skill level” objectively says until you get back to old Rank.
NOW, if you take all of that and go back to the original point…
If you’re playing against a huge pool of players at a given MMR before you de-rank, you’re playing against the same pool of players AFTER you de-rank…and if you were winning 50% of the time before…it follows you should still win 50% of the time at that MMR.
If they are doing as you suggest then they are inflating Rank gains artificially regardless of MMR. Given we can only see a change at best every 7 games, that makes it tough to measure/verify.
But you can see yourself by the gains or lack thereof if you went 7-7, 7-9, 7-2, etc. it’s not that hard if you play a lot of ranked.
We absolutely know the decay is to make things feel better as they’ve said as much.
We do know that SR DOES update after every game because Top 500 rankings update after every game and your last rank at the end of the season is based on the last games played and not just the last update. So you can have your last update as Masters 4 and have your final rank as Masters 5.
You’ve lost me mate - I don’t actually know what you as disagreeing with anymore.
I don’t understand this mate. Not trying to be difficult but do you mind rephrasing? I’m having trouble getting what exactly you are disagreeing with.
Please note that they have specifically said that SR “chases” MMR and is a more “digestible” number than MMR in the links I shared earlier.
To simplify what I believe, what I have experienced and what my friends and streamers have experienced:
- if you play a lot, you will stabilise at a rank where you have 50% win rate… let’s say masters 5. If you have played enough, your SR is a fair reflection of your MMR. That is to say you are getting 50% matches with people on average that have the same rank.
- at the end of the season, you are deranked without your MMR changing… let’s say to plat 2
-you will still be in the same matches with the same people - with a 50% win rate your rank will sharply increase - you go 7-7 and your next update is to Diamond 5. You go 7-9 and your next update is to Diamond 4. You go 7-6 and your next update is to Diamond 2.
- this continues until your get to Masters 5 UNLESS there is a huge change in the skill level of the entire playerbase relative to your own. In that case but if that was the case, that should be reflected in your results and thus your MMR would change.
No - I am saying they are artificially DEFLATING your rank during your placements so that you are guaranteed SR increase even if your MMR drops to give you a sense of progress even if you aren’t really improving AND to give everyone a reason to play comp every season.
1 Like
Let’s focus on just these two items. This really covers everything I’ve been whining about.
You’re suggesting that the same players with the same MMR playing the same games and winning 50% of the time will yield negligible Rank on a Sunday but give you the equivalent of ~300 SR in 7 wins on a Tuesday right after a de-rank?
Can you see why I had the observation that I had?
I’ve read the MMR doc, the Initialize OW2 doc, the Season 2 Update doc and your archived Kaplan comments, and I still see nothing that suggests that the implementation you describe exists.
I think the actuals are more likely to be due to the volatile state of the game, the volatile changes to population, and the common reports that people are seeing wild matchmaking.
Anyways, I’ll go back to the quote you already cited from the most reason Season 2 update…
The only way I can interpret that is “true skill level” is defined by the point at which you’re not able to consistently beat opponents that have a higher MMR more than 50% of the time.
Then you are suggesting that the idea SR chases MMR has changed between OW1 and OW2.
Exactly. That is what is happening. If you can’t use your own games and games of friends you partner with to see this, look at top 500 streamers that deranked to low masters and see what their updates looked like.
But none of those changes from Sunday to Tuesday. The matchmaking is a wild experience but the logic is constant. The match maker is still trying to find a fair game so there isn’t really a reason to see such wild variance.
I understand what you are getting at but the logic simply doesn’t make sense to me.
Why create your own definition when we have already been given one?
When a player wins more than 50% of their matches, we start to increase their MMR faster. This levels out once we see the player’s MMR value has them joining fair matches, which corresponds to a 50% win-rate.
Which I understand that the result is the same but the way you worded it suggests that you are matched with people with a higher MMR than you deliberately when there is nothing stating that. Simply that it keeps trying to find you fair games with your MMR.
I have had very similar experiences for the past 6 years in OW1 to what I’m seeing in OW2. The only difference for me has been the rank reset.
Your suggestion that the volatility in the player base accounts for the massive differences in rank gain at the start and end of the season is our core difference and I honestly think you haven’t seen the reality of what is happening and are listening to a few too many blind forum anecdotes.
I was talking about the INFLATED Rank gains (in response to your rightly noted DEFLATED Rank) specifically by this. I already stated that I agree that MMR change impacts Rank change.
And when I read the available information, I wasn’t able to find a reason. There’s “shoulds” and “player skill” references but not a satisfying clear statement. Regarding, the top 3% of the player base, I really don’t think it’s reasonable to measure games systems by them. I peaked in mid Diamond and spent most of my time Plat. My digging into this argument is part of deciding if I want to get back into comp. So, I definitely have to concede lack of personal recent experience in Comp. I also hate to admit that 100% of my gaming network quit OW years ago. The handful that tried OW2 quit after a few weeks into season 1.
On the contrary I think the beginning of a new season is the most volatile period to pique the communities’ interest and engagement. I think the end of a season is the least engaging. Completely anecdotal.
Yeah, close enough to the same thing. What I was really getting at though is “true skill level” isn’t last season’s Skill Tier & Division. I just can’t find anything that says a historic Rank is a variable and calling out that even Blizzard uses “rank up to your true skill level”. Why multiple terms Blizzard?
I also believe Blizzard when they say “try to find a fair match”. Which implies that you also get less than fair matches which should impact Rank adjustments respectively.
Thanks for this. The smell and eye test means a lot. Last question, what is your rank?
But you did not state that a large delta between your actual MMR and the MMR that your rank should represent, creates large gains per win as the SR “chases” your MMR. That’s the core of this.
The point is that your SR doesn’t really change at the end of the season if you’ve played a lot and changes dramatically at the start of the season.
There is nothing to state that there is any MMR disparity between teams that is classed as material enough by the matchmaker to award people huge amounts of SR for wins.
Volatility mentioned by Blizzard is based on new players coming back and not having an accurate MMR, not having an MMR disparity between teams.
They mentioned this is a minor (in bold) factor in SR adjustments in the archive I linked, with emphasis on the minor - from personal experience, it is not even close to the impact of very poor or very good performance in a match.
Before getting into an argument like this, you should probably should have mentioned this - it would have saved both of us a lot of time 
.
I’ve been talking about this from the perspective of someone that plays comp regularly so the SR is fairly well adjusted to the MMR.
If you haven’t played for a while then the situation will be very different.
If your historic rank wasn’t last season, then I agree completely. Also, your MMR is likely not accurate. They have decayed your MMR to account for this if you didn’t play season 1 but it ls likely it’s not accurate - your “true” skill level could be higher or lower your current MMR.
Different ranks for different roles but Masters now for tank.
it is a bell curve, of course it is relative to the papulation. and you said that it wasnt relevant to the conversation. that is why i called you a prick. because it was.
and while there is an Equivalent MMR to certain ranks, it is not a hard line, not did i say it was a hard line, just that there is an equivalency.
i mention SR as it is the only thing we see. and i explain how it can show approximately what your rank is, but only after certain number of games.
if MMR is correct for all 10 players in a match, and if all 10 players are of similar MMR, what do you expect is the chance of either side winning? i would say about 50%.
therefore, if a player is above 50% winrate they are below their correct MMR. and if they are below 50% winrate they are above their correct MMR and have the be adjusted accordingly. notice at no point are you forcing a winrate. just adjusting the players that are being matched so that everyone is near 50% winrate when all is said and done.
If matches are meant to be fair, doesn’t that just mean that whether you’re in Silver or Diamond, because everyone is always supposed to be of equal skill to you, it makes it inherently super difficult to climb? To the point your Gold match feels like you’re in a Diamond match? Good players would climb out if they’re far above average, but it’s also going to stick a lot of people where they really shouldn’t be, which also results in awkward games that feel imbalanced. I’d imagine it’s even worse if a group of people play together and they all have varying degrees of skill but are in the relatively same bracket.
Yeah, true. This is because I disagree with the notion of “MMR that your rank should represent”. We’ve already beat that idea up quite a bit and you believe me to be blatantly wrong, which is fine.
And my point is that it is reasonable to believe that the player pool between that Sunday and Tuesday could be wildly different. Also like I said entirely anecdotal.
For example, it is reasonable to believe the pool of players has dwindled down some at the end of the season. Those that are still active are fresh. On Tuesday, new season, those that haven’t played for a few weeks but are similar MMR come back. Perhaps get blown out against people that are fresh at the same MMR and reasonably those winners win even a greater share than normal rank gain, etc. Speculation at best.
Agreed, for your rank, new players shouldn’t be a variable per resent Blizzard updates.
Fair. 
Edit: Though I should mention, I do have like 400+ hours in Comp, I just haven’t done so recently. The reasons I haven’t is because I’m trying to understand these changes and I had wanted to wait for the initial complaints to die down (though that’s not happening).
I’m not talking about myself here but in general. The notion of “true skill” which Blizzard seems to be using is not the same as Skill Tier & Division. Hence historic rank is a false notion. But again, we’ve gone back and forth and you believe I’m wrong, which is fine.
I have to wonder how much being on the top 5% of the player base impacts things. Hard to know. You can imagine the MMR range in matches shrinks significantly at the Gold/Plat area which should cover more like 50% of the player base. Yet outside of a couple of exceptions, everybody is beholden to the same engine.
Other than calling me a prick, which was uncalled for, it sounds like we’re not far off then on this item.
That’s fair.
I was responding to your example match where the MMR was wildly different. Which shouldn’t happen in the first case if the goal is 50% win rate. In your example you said the equivalent of Plat 2 vs the equivalent of Diamond 3.
But 100% agree with your response there.
What if? Everything except MMR is deleted, including skill tiers, divisions, SR, and competitive queues.
We merge the MMR from QP and Comp.
Show MMR as a ranking number. They can choose the number they want (from 1 to 10,000, for example).
You have a better pool of players with everyone united in the same queue, and at the same time everyone can compete in Quick Play.
Problem solved (I realize it’s overly simplistic, but it might work)
Yes it’s our core disagreement but I have trouble understanding why you disagree.
To me it’s pretty clear:
- if SR chases MMR and
- there is a corresponding expected SR to your current MMR and
- those converge with enough time if your MMR is constant
Then them bringing the language of “true skill” and “skill tiers” makes perfect sense. You’ve just chosen not to believe all that even though it fits together.
You’ve instead chosen to believe that Blizzard are somehow wrong or confused with their language. For what reason and on what basis, I don’t really understand.
Well I initially placed in silver in both support and DPS and I’m climbing likely to diamond in DPS although I’m plat in both and I’ve experienced the same thing really.
After you’ve actually played comp in OW2, I’m sure your opinion will change.
fiction. once upon a time win streaks were more common before implementation of artificial outcomes which began 2019 and gradually became more artificially difficult to achieve win streaks due to implementation of matchmaking manipulation. when you rig-n-watch players leave-n-quit. simple math.
1 Like
Sounds like your just going to ride it out and see how it settles. Unless you get a dramatic drop in the player base or revenue from the game.
Balancing OverWatch is an almost impossible task anyway. You cant account for which class or Hero a player will pick. Skill in each different class will vary and skill with each individual Hero. There are to many Heroes with different skill ceilings, there are to many variables to balance.