Groups and Matchmaking in Overwatch

[Kaedi-11739]

I think it was on a video, like from the first 6 months overwatch was out. maybe it was a dream lol

I will check the DUVs.

[Kaawumba-1133]

@RexOverwatch-1555

I remember quite a while ago Jeff said that figuring out the SR of a group is not as simple as adding up the SRs and dividing by the amount of players.

So, how is group SR calculated. Because if group SR is different from the average SR of all the group members, that just seems wrong, thank you.

I think you are referring to this post:

“For matchmaking groups, there’s actually two separate issues that we try to solve. The first issue is ‘How do we handle groups formed of players with different MMRs?’. With season 2 we prevented players of REALLY disparate Skill Rating from grouping, but there’s still some variance we need to handle. Over time we’ve tested different models to try and see what’s best and are now using what tested most accurately. (Hint: it’s not simply averaging the MMRs)”

– Scott Mercer Overwatch Forums

Much of that post is now obsolete, so it has to be read very carefully, if at all.

However, what I expect to still be the case is that a group of 2000 + five 1000s is not considered to be equal to a group of six 1166s. It would not be predicted to be a 50/50 match. I’d want a fresh confirmation from Scott or Jeff to be sure.

What Scott’s post from two days ago indicates is that it doesn’t make a difference if that group of 2000 + five 1000s grouped intentionally or was thrown together by the matchmaker.

[Kaawumba-1133]

However fellow power user, Kaawumba, does have it documented in this SR guide that group size currently have no effect toward determining the team SR.

That is purely because of Scott’s post, quoted below. I edited my guide last night.

Let me rephrase something I noted in the post:

Group size doesn’t matter for the purposes of calculating expected win chance for a game, what matters is the SR of the players on each team. Since grouping status doesn’t change the expected win chance, it doesn’t affect the amount of SR change you experience after a win or loss.

[RexOverwatch-1555]

(Hint: it’s not simply averaging the MMRs)

yup that’s it! thanks kaawumba.

If you want to find the correct SR of a group, you must take the average. Any other method, and I’m going to #TeamCuthbert.

If they explained how they calculate group SR, I’d have to take it on a case by case basis.

And as an after thought, no matter which method they used to group calculate SR, I’m assuming they use the same method whether the team is a 6 stack or 6 solo querers.

[RexOverwatch-1555]

here is my speculation of an way they might calculate team SR that isn’t simply an average.

So they’re saying that you need to play regularly to have a accurate SR. So if some people on the team haven’t played in a while, then they won’t necessarily contribute one-sixth of the their SR towards the total for the group. Seems kinda logical…

[Steadfast-11644]

Did you fix the placement matches? I have been gold before but every time now that I do my placements I am always bronze, and I am not a trash player. I should be able to place higher if, for example, I win all ten games, and I shouldn’t get punished for what happened last season. Also, if I win all my placement matches, and get put in bronze, I am leaving this game. Not putting up with players that don’t know how to move forward.

[Kaawumba-1133]

I suspect that they throw every match every played into a machine learning algorithm, where only the current MMRs and the game result are input. From this, they can predict a win likelihood for any given match given the MMRs of the two teams. This would be in no way rigging, a la Cuthbert.

And that they only take into account recent inactivity when determining how much to move someone’s MMR/SR after the match.

But unless they say what they are doing, we are just speculating.

[Kaawumba-1133]

Did you fix the placement matches? I have been gold before but every time now that I do my placements I am always bronze, and I am not a trash player. I should be able to place higher if, for example, I win all ten games, and I shouldn’t get punished for what happened last season. Also, if I win all my placement matches, and get put in bronze, I am leaving this game. Not putting up with players that don’t know how to move forward.

You have to play during the season and work your way out of bronze. Season placements are dominated by previous season end.

[Arbacus-11537]

my point is their design goals they talk about here are out of touch with the goal of making competitive mode in overwatch competitive in the same way you would see in pro gameplay only less skilled obviously.

All that was discussed here is only part of it none of the rest of the details in how they calculate the as close to 50 % chance to win for each team is discussed here other than party size. There are other factors obviously. Specifically it was noticed that the matchmaker defintiely on the eye test makes it easier to group up now than before in terms of punishment players notice in their SR hits compared with solo or smaller party play at the asme SR. All that was said here again is a smokescreen about the parts that didnt change like you can use the same forumula for SR calculation but have the matchmaker pick different players because it now thinks of the win %tage different. So what changed isnt the forumula for SR its something related to party play in the math of how they arrive at that 50 % chance to win(and again obviously its not that they changed it wanting to match equal sized party as explained). Process of ellimination leaves thigns like did they have a more flexible hidden MMR with boosts if you party up(which adds weight on how strong the nemy players have to be to get to that 50 % win chance) or something that got either narrowed or the party boost removed thats the kind of thing which would have changed.

[Serenus-11584]

In OW, there is no such thing as being the “best” of a category. You know like in real sports when you are put into a division and if you re better, then you have something like 80 or 90% winrate. So you have a goal, you can train and being rewarded with victories and ultimately win the division and gain the right to go to an harder one.

If you want to enjoy 80% - 90% win rate when you are the “best” of a category, you must also be willing to accept 10% - 20% win rate when you are the “worst” of a category. I think you will be hard pressed to find players who are fine with the latter.

[Kaedi-11739]

Well I am leaving this game. No point of playing if im just going to be placed in bronze every season. Blizzard used to make great competitive games. Now they are just shooting themselves in the head.

To be honest, I once fell to Bronze in Season 3, and easily carried myself out by the end of season 4. It just takes a lot of Pharah Rockets… lots and lots of Pharah rockets.

[NepaliBabu-11709]

have u ever cared of giving a thought that maybe you are in bronze rank all the time cuz u belong there and simply are not good enough to be in higher tier???

[Hanzomain94-2958]

Would have been better if you gain more SR if your predicted chance of winning was >50% and gain less if it’s <50% since you’re beating the odds. Your system is unfair, doesn’t reward hard work and favors the better players

[Hanzomain94-2958]

What do you want them to do if you are bad at the game?

[Kaawumba-1133]

Would have been better if you gain more SR if your predicted chance of winning was >50% and gain less if it’s <50% since you’re beating the odds. Your system is unfair, doesn’t reward hard work and favors the better players

You’re backwards. A player who is expected to win gains less because it was expected. Imagine if a player was expected to have 100% chance to win. When he wins, he shouldn’t gain anything because the result was expected.

[Hanzomain94-2958]

No you are. If you are expected to win against me, you’ll gain the SR anyway. The favor is on your side. But if I beat you then I’m the underdog and should gain more because I beat the preferred. Its basically just rewarding you for being the better player and giving me no reward at all. You’ll win 20 SR for winning the match, +5 SR for winning and being the preferred. How does that help me at all?

[Kaawumba-1133]

But if I beat you then I’m the underdog and should gain more because I beat the preferred.

This is how it works. It appears you did not write what you intended in your first post.

[Kasp-21722]

What about the bots only saying stunning in 2 and gather up to push with rally. They’re same sr with the rest

[MHz-11115]

My friend, I would like to discuss how terrible the matchmaker design is. I know what you are doing. You rig matches for players to move up or down based on performance, but you do not understand how bad this is. The metrics you are using in order to evaluate individual performance are not entirely what a good player is. There are many players who are contributing and playing well but are not being rewarded because of this matchmaker system.

I do not think you understand competitive or you would know that this is bad.

[JuicyBrucie-1683]

Cool so they finally admit the force 50%…where are all those fanboys calling us conspiracy nuts