School me on SR win/loss adjustments

Competitive Discussion
1

So in general, what does it mean if your SR gains from wins tend to be much greater than your SR losses from losing? And vice versa?

And is it scaled differently depending on your skill tier?

Background: I’ve been in bronze pretty much since launch. I’ve fluctuated between 900 and 1200 at the last few season’s ends, but my SR losses were usually greater than my wins, which didn’t really matter to me as long as I kept my win rate around 60%, because I was still (very) slowly climbing.

But I went through a HUGE losing streak early in this season, and dropped all the way down to near 700.

Then this weekend, I started winning again. Over the weekend, I won 10 matches and lost 4. And I noticed that my SR sometimes went up by as much as 38 SR for a single win. I’m using getting 21-22 for a win.

And the losses were similarly blunted. I’m used to losing 23-25 for a loss, but this weekend it was more like 15-18 per loss.

So I climbed about 200 SR this weekend, on a net of 6 wins. That’s an average of 33 SR gain for those wins.

So were the increased gains and reduced losses because I dipped so low? I’ve been that low before and didn’t experience that. Or is it because my hidden MMR is higher that my SR, and the system is trying to bump me up to where it thinks I belong? Or am I just outperforming the typical bronze level players and PBSR is kicking in and I’m maybe even (gasp!) starting to climb in earnest?

2

This weekend is when the smurf wave has ended, which was caused by two game sales in a row.
And yes - you fell too low, and matchmaker has recognized that, and gave you SR bonus for wins and subtracted less SR for losses.

1 Like
3

There is a point near the very bottom where the game is fixed to prevent you from dropping much further and gives you more SR for wins than loses.
I have never been that low so I don’t know where it starts.

However the bigger factor is that the stats for an average player in bronze are not very good, so even a small improvement results in the a large jump in MMR and hence a significant boost in SR gains.

With respect to performance you don’t have to do much - just focus on not dying this will likely have the greatest effect on your K:D ratio which seems to be the primary factor in SR gains in bronze.

4

The truth is we don’t fully know for sure what the SR system is doing.

We’ve been told a few things and using the power of deduction, we’ve filled in most of the gaps. Even with that though there are some things which still don’t make sense.

Unfortunately Blizzard are unlikely to ever tell us the full picture, as they believe that information could be used to game the system, although take that with a pinch of salt. I think it’s more likely they don’t want us to know as it’ll result in a metric tonne of angry posts about why the system works in a particular way.

What we do know:

SR adjustments seem to have a baseline value of +/- 25.

Below diamond, there is Performance Based Skill Rating (PBSR), which then acts as a multiplier on top of that (+/-).
Blizzard have told us that PBSR is a calculation which compares the individuals stats to that of an average player within that rank on that hero on that map.

Average is defined as the mean value within a given rank.

They haven’t said what they do with this, but it seems reasonable that it’s a sliding scale up and down depending where you compare to that player.

We believe they use MMR for this, but who knows.

The result is an adjustment on top of the baseline.
If you play well you get more, play less well you get less.

In addition to the baseline and PBSR (below diamond), there is then an adjustment made depending on the bias between the two competing teams.
What I mean by this is that the team that statistically should win (based on MMR) will earn less SR if they do win, as the match was somewhat biased.

The opposite applies if team that the system calculated should have won, lost. They lose a lot.

If your team is calculated as less chance of winning, but you did win, you get more.
This correction they believe levels out the situation of unbalanced teams.

The biggest issue I have with things like this, is how they calculate probability is terrible.
They don’t consider what hero’s people ar likely to play. I can’t tell you how many times I’ve had matches with 4+ mercy mains, but the system doesn’t care, it sees the players as people performing within their SR bracket, not what hero’s / role they play.

This is the age old debate about role queue.
Role queue in my opinion isn’t just about balancing teams, it’s about balancing the whole system, which in my opinion is fundamentally flawed.

However, after thousands of games, I’m confident the system is not rigged, it’s just bad.

An example of something which still doesn’t make sense:
I was playing in masters this weekend with two friends. We’re all very similar SR, all in masters, therefore no PBSR.
One match I gained 22 SR, one mate got 28 SR and the other got 24.
How does this even make any sense with no PBSR?

SR should be flat wins and losses across the team, based on what we’ve been told and think we understanding.
How can team bias be applied differently to different players in the team, that makes no sense. I fully believe they have NOT fully removed PBSR and there is likely a bug in the system.

Hope that helps.

1 Like
5

I wondered about this. I have felt that there was YUUUUGE wave of smurfs and deliberate de-rankers that started when Season 12 started, far bigger than I have ever seen before in any other season. But everyone on the forums has just been like, “No, if you lose it’s just because you suck and for no other reason.”

I dunno. I’m sure I do suck, but it really isn’t helpful for climbing when there are GMs and masters and even Top500 players crudding up the mess that bronze already is. It’s been like 1 or 2 out of every 5 matches until this weekend. I even played a team that admitted to being normally at 4800SR.

1 Like
6

The game compares your performance (e:d ratio seems to weight heavily for DPS) with players at your rank in the same match.

If you have better stats you gain bonus SR (more Sr on win less Sr on loss) and vice-versa.

There are two schools of thought for ranking up:

1- fill.
You let your teammates use their best heroes, this gives you more wins at less SR per win (this is what you seem to be doing)

2- main/one tricking
You get better at one character, and become better than average. You win less matches (since you force others to fill and not play their best) but since you are specialized you lose less Sr on loses and gain more on wins.

Personally I’m on camp 2. I like to be good at my hero, actually I just play to kill people and show them I’m better than them #lolissues

7

Funny you say that, because up until this weekend, I WAS filling a lot. But this weekend I played nearly every match as Rein (with a round or two as Orisa, and I think even Lucio one round).

I had been getting bored with Rein, but lately I’ve been really enjoying playing him. It’s like I found some new aspect of his tankiness to appreciate. Had a blast and maybe (?) it led to bigger SR gains? It’s not the first time I’ve played Rein, but it is the first time I’ve seen gains like this.

1 Like
8

It’s called gitting gud :slight_smile:

9

4800 sr… have you ever looked at the top 500 leader boards out of curiosity? Number 1 normally hovers about 4600-4700 and that is only a few handful who get that high then like the other 450 never pass 4400 lol. Yes it makes perfect sense the best players in the world want to spend their time ruining your game in bronze.

10

Well, sure, they could have been lying. But it’s not that surprising either way. There’s an entire cadre of people who only find fun in this game by ruining it for others, and what games are easier to ruin than bronze? I mean, you’ve been around since season 1, so you probably knew that already, right?

I’ve had a 4200+ ON MY TEAM in bronze just in the last month through LFG, he played Widow and pretty much killed anyone he wanted for the entire match.

I mean, why would ANYONE want to spend ANY amount of time ruining bronze games? But it happens. A lot. Even if they were lying about being at 4800, that’s not really the point, is it? Maybe he was 3800, but that doesn’t make it any more appropriate, or make any more sense, for someone who’s at 3800 to ruin bronze matches than it does for someone at 4800. This wasn’t smurfing. They weren’t trying to win, and they never stepped on the point (except for the second round, right after we capped it, 30+ minutes in). They were trying to drag the game out so long that we left and took a leaver penalty. But none of us did. The match went 90 minutes, mostly us getting spawn camped.

What I can say with absolute certainty is that I reported all 6 of them, and got notice back from Blizzard that they took action. So yeah. They didn’t belong in bronze. That was the point, ICYMI.

11

Because while there is no PBSR there is still MMR. You might all have very similar SR, but your MMRs could differ quite a bit. The system still adjusts the amount of SR gained/lossed based on the difference between your SR and MMR.

However, typically SR follows MMR quite closely so this has a small affect on SR. But you guys are in Master so maybe some people had decayed and their MMR was higher than the SR by quite a lot?

12

Nah, don’t listen to the forum answers (mine included). You should also do not listen to braggings in game chat.
Players in this game are incredibly biased. Those, who reached diamond and above, relate this fact to them beinggud (which is so rare actually, that it’s closer to non-existent). Those, who are stuck below diamond, blame teammates and matchmaker (which is true, but to a certain degree).

But I’ve recorded my gameplay stats for several seasons. And this goes from observable data: once the game goes on sale, I lose 200-500 SR within a season, only to get it back, once another smurf wave is correctly placed. I has been like this for 4 seasons already.

13

“However, after thousands of games, I’m confident the system is not rigged, it’s just bad.”

The most accurate and succinct comment on the MMR/SR system I have ever read.

14

These are two possibilities. I expect that the second is more likely, but I can’t actually prove it. Another possibility is that because your rank is so low, you are expected to lose most games, so you win more SR on on a win than you lose on a loss.

I have a longer explainer at How Competitive Skill Rating Works (Season 12).

15

I was thinking it might be MMR-related, because since I started losing like crazy in comp at the start of the season, I supplemented with lots of QP and FFA matches. Did MUCH better in those than comp, so if winning non-comp games makes your hidden MMR go up, that could be a factor, no?

16

Wow, 90 minute game? That’s crazy.

17

No. “Long story short, you have a separate matchmaking rating for Competitive Play than you do for other modes” – Jeff Kaplan (Overwatch Forums)

18

We all have our own experiences, some people on this forum believe MMR and the SR system is completely rigged and predetermined.

I’ve not seen any such evidence, in my experience of my games.

Most of the bad games are from imbalance of the match due to lack of role select.
Practically every match has 3+ support main, myself included and I don’t play support because I’m good at other roles, in fact I’m pretty bad at dps and average at best as rank.

Hence my comment, it’s not rigged, it’s just bad.