You have to break it down even more, how does a team get created?
does it “try” or is it based on time?
1 Like
Why must I do that? Every time you queue into a game, at the starting screen, take a look at both team’s average SR. They will be similar.
I don’t know how specific players are matched into a team. I only know the matchmaker uses an adaptation of the Elo system, in which a game can be determined to be balanced based on the rating of the players. In this case, rating is SR.
For your last question, I don’t know what you mean by “try.” I also don’t know how time factors in, but I would guess that if a balanced match can’t be made in some period of time, the criteria will be loosened and a slightly less balanced match would be made. But unbalanced matches are allowed in the Elo system - rating adjustments are affected by how balanced the match is. For example, if a match was made where team A has an average SR of <500, while team B has an average SR of 4500, then team A is almost certain to lose and they will not lose much SR. However, if team A did win, they would be rewarded a massive amount of SR.
The game itself is very relevant to the discussion, though. Again, if there’s a focus on single player gameplay, you can’t blame teammates for holding you back. Overwatch has always prided itself in being team focused, though, and more so than other games. It makes sense that it’d see more complaints about teams not being as… ideal as one might hope.
This may be something along the lines of what you’re looking for:
https://mystgraphics.com/overwatchforumarchive/competitive-howMMRworks.html#post-3
When the matchmaker creates a match, it determines the % chance for each team to win based on the match it made. The VAST majority of matches are usually near to 50% (especially if you’re a player closer to median skill rating and you’re not in a group), but I’ve definitely seen logs of matches where that’s really not the case and my eyebrows raise.
The unfortunate truth is that there is not always a “perfect” match for you, especially at very high (and very low!) skill ratings where there’s fewer players of similar skill. Then you throw in the desire to match groups vs. groups, with everybody having low latency, and doing ALL of this as fast as possible even though it’s the 3AM offpeak… it can get tough. We’ve tried different tunings with regards to wait times, and the improvements were unfortunately modest as we increased the time to wait. Still, this is an area we’re always looking to improve and tune better.
Fortunately, when we do put you in a match that we know isn’t a 50/50, we adjust your SR gain or loss based on your calculated change of winning. So if you did get placed into a match with only a 20% chance to win and then you lose, you shouldn’t lose much SR.
Not necessarily.
A big reason we have SR and MMR remains hidden is because it isn’t fun to look at. It may stay the same despite winning or losing if certain factors line up appropriately, and let’s be real, it’s a lot more fun to see SR climb after a win than it would be to see your MMR stay static.
MMR works very similarly to SR. There are some minor differences that make it feel worse though, when you just watch that number. For example, it’s possible to win a match and not gain any MMR. We make it so that if you win a match, you always gain SR – even if it’s just a little bit – to feel psychologically rewarding. But MMR’s entire goal is creating fair matches – which isn’t always fun to look at and certainly not “rewarding” for players looking for pats on the back or a sense of progression. So SR “chases” your MMR very closely, except in a rare case of severe SR decay at GM/Masters/Diamond level of play.
Jeff also went into good detail about how matchmaking works at a high level and why some people just aren’t satisfied with it, even when it works properly. (It is quite the read, though, and you do not want me to quote it all here)
https://mystgraphics.com/overwatchforumarchive/competitive-matchmakingandavoidthisplayerremoved.html
And as a side note, if you want to try and look up this sort of stuff, WyomingMyst has done a great job at keeping an updated archive, which is especially useful now that the old forums have been deleted
2 Likes
yea it sure is forged dude. I just had a game where 2 of my teammates were border gold…and i am mid plat like WTF? How does this even work? Of course I lost that game…this game is BS, calling it right here:)
1 Like
Dev post is on the old forum. I dont have time to find it, but it was said there that matchmaker is trying to create 50/50 games, meaning he will find two teams who have same chance to win. As devs said, in worst case scenerio is match 40/60, but matchmaker will never go lower with chance for one team. 40/60 is caused usualy by very small player base.
My original thing I said, which you said was not correct, the system tries to predict your odds of winning. He never says “how” just that it DOES try to predict.
Here are some quotes I found intersting.
“We do need to do a better job of not placing players into “unwinnable” matches.”
- 2016 post still happening today.
“When the matchmaker creates a match, it determines the % chance for each team to win based on the match it made.”
THIS IS PREDICTION!!! But on the team level? but how can it not trickle down to a player level?
“The VAST majority of matches are usually near to 50% (especially if you’re a player closer to median skill rating and you’re not in a group), but I’ve definitely seen logs of matches where that’s really not the case and my eyebrows raise.”
So how are the logs now? Maybe the roflstomps are one offs? we need to know the sample and population. He admitted they have logs so - what are they now?
“The unfortunate truth is that there is not always a “perfect” match for you, especially at very high (and very low!) skill ratings where there’s fewer players of similar skill.”
I made a post about time of day… this seems to confirm that feeling.
“Then you throw in the desire to match groups vs. groups, with everybody having low latency, and doing ALL of this as fast as possible even though it’s the 3AM offpeak… it can get tough.”
Low computational load and algorithm, but again what is the algorithm? we’ll never know, but I have a feeling win/loss deriative is in there.
“Fortunately, when we do put you in a match that we know isn’t a 50/50, we adjust your SR gain or loss based on your calculated change of winning.”
Again pre-calculated chance of winning - a.k.a prediction
“So if you did get placed into a match with only a 20% chance to win and then you lose, you shouldn’t lose much SR.”
Prediction but soften the blow?
Why would system shoud do that? There is literaly no reason for it. You are really confused with terms. Matchmaker is trying to predict TEAM CHANCE not individual chance of players.
Regardless of how or what, the outcome is its failing.
How does it make up a team? Current SR? Current MMR?
Well then it goes back to my other posts, your journey to that current state could vary, and the system is not accounting of that.
Honestly, its impossible. Its like tracking molecules and their history trying to figure out where are they going to go.
I think they should just go the random route.
1 Like
MMR, matchmaker take MMR of players and combine it to get even teams.
MMR is a derivative of win/loss but at a “rate” of winning and losing.
So again, it never says how the teams are made, but if people have been losing, does it match you with winners? Would you want to be with a team of losers?
Maybe you won cause of last game’s smurf.
1 Like
Try to write down timezones you play.
Dony play the time/days u r on loosing streak.
Player base ( babies , drunks etc play a big part)
And if u play early in the morning with your coffe like me u end up with players from bronze to plat as there are not enouph players
Also stop playing after 2 loses ( or go arcade) chances are u are tired and on auto pilot
1 Like
Your MMR has nothing to do with winning or losing. MMR is raw game skill, mostly game sense+mechanical.
Yes they said how are teams made, read the stuff above from the links. They take players and create teams using MMR so they have same chance to win in best possible scenerio.
Post this in time to play.
By all means, if say MMR is based on those medals then I see why its all messed up.
2 Likes
Sorry I got too much time on my hands.
I did go back and read, this is what Jeff said.
“MMR works very similarly to SR. There are some minor differences that make it feel worse though, when you just watch that number.”
Raw game skills does not sound like similar to SR but minor difference.
MMR feels like that prediction element they talked about. It only goes up if they thought you would lose.
1 Like
No, MMR goes up if you overperform players playing your hero in same rank.
They designed the system on psychological principles of what players would “enjoy” instead of creating a system that would be “fair”. Until this is changed and blizzard believes the game is fun enough stand alone then a MM that tries to create psychological positive outcomes will always be present and oppress integrity.
Ahh yes, they compare the 10 min average.
I remember not liking that either, because any type of stat is based on your team, enemy team and map.
You might have a career low just cause.
1 Like
It’s really ballsy to suggest the games are ‘rigged’ on this forum.
However, if one can substantiate their claims they might be believed. I’m going to link a topic which may elucidate what’s happening.
The system predicts your odds of winning based on your team’s average SR.
The approximate formula based on the Elo rating system is (and I’m sorry if it looks terrible - these forums don’t have any mathematical typesetting available):
E(A) = 1/(1 + 10^(-D/832))
Where D is the difference between team A’s SR and team B’s SR. So D=SR(A)-SR(B). The 832 refers to how steeply the change in rating corresponds to a change in skill. The typical Elo system uses a value of 400, but by a statistical analysis of over 3000 competitive games, it appears Overwatch uses a value near 832.