Groups and Matchmaking in Overwatch


(Scott Mercer) #1

The Looking for Group tool in 1.25 is an exciting new in-game feature that gives players more control over their online play experience in Overwatch. It allows players to find like-minded or similarly skilled individuals who want to work together. Groups can lead to better team play with less negativity and, ultimately, more fun. Since Overwatch is at its very core a team game, there’s really no better way to play.

We’ve seen some reluctance from the community when it comes to grouping, especially in Competitive Play. There are a lot of misconceptions about how grouping is handled by the matchmaker, as well as how grouping affects Skill Rating changes. I want to address some of these by explaining in detail how our matchmaking system handles groups, and also examine the effects of grouping on win rates. I’ll include quite a bit of data and statistics to help with the explanations.

The first and maybe most common misconception I want to correct is the belief that the Competitive Play system decreases your SR gains and increases your SR losses when playing in a group. The simple answer here is that there’s no SR penalty based on your group status. It doesn’t matter if you’re solo, grouped with one other person, or in a full group of six. If you are a 2800 SR player grouped with five other 2800 SR players, the SR change after a win or loss is the exact same as if you played the game solo with five other solo 2800 SR players. We also do not artificially inflate the SR of the players in a group when finding matches. There is simply no penalty at all for the purposes of calculating SR and matchmaking.

There are many other reasons for the SR gained or lost after a game to differ in magnitude that have nothing to do with grouping:

  • What was the quality of the enemy team and your predicted win chance? You gain more on a win if your predicted win chance was <50%. You gain less on a win if your predicted win chance was >50%.
  • Are you playing consistently? New accounts or accounts that have been inactive will see larger magnitude changes both upwards and downwards. This settles back to normal as you play additional games.
  • Did your 3000+ SR recently decay due to inactivity? If this happened, you’ll gain more SR on a win until you get back to your “undecayed” SR.
  • Are you a Platinum-tier player or below? If you performed particularly well or worse than what is considered a typical performance during a match, then there’s a small SR modification to reflect that.
  • Is your current SR really high? Your SR increases less on a win than it decreases on a loss as your SR approaches the systemic limit of 5000.

Now let’s talk about groups and matchmaking a bit. Interestingly enough, there’s actually more grouping going on than you might think! For the next set of data, we’re going to look at all competitive games from February 1st up to May 28th of this year.

  • Only 16% of all games purely consisted of solo players.
  • The most common match composition is 2,1,1,1,1 vs. 2,1,1,1,1, which represents 28% of the all matches.
  • If you are in a full team of solo players, 73% of those matches were against another team of solo players and another 24% were against a 2,1,1,1,1 team composition. Only 3% of the enemy teams were groups of three or more players.
  • If you’re in a duo, 74% of your matches are against other 2,1,1,1,1 teams, 14% against 1,1,1,1,1,1 teams, and 8% against 3,1,1,1 teams. Only 4% of the enemy teams you faced had groups of four players or more.
  • If you’re in a full six player group, 92% of your matches are against the combination of other six player groups; 5,1 teams; 4,2 teams; 3,3 teams; and 2,2,2 teams.

The matchmaker was designed to try and create games with equivalent-sized groups, especially for solo players, and these statistics show that it’s doing a pretty decent job. We believe that games with equivalent groups create the fairest possible experience, and fair games create the best chance of players having a fun experience. When the matchmaker does compromise, it’s usually during off-hours when there aren’t as many available players, as well as at the lowest and highest extremes of SR.

There are limits to how much we compromise, however.

Several months back we implemented a restriction that prevents the matchmaker from creating games that are severely one-sided. Whenever it wants to make a game, it calculates an expected win percentage for each team based on the SR of all the players. If one of the teams doesn’t have at least a predicted 40% win rate, then we simply don’t create the game. Even if you do end up in game where your predicted win rate is close to 40%, it’s important to remember that it also means that your SR will drop less when you lose, and you will gain more SR if you win. To help explain this further, here’s a simplified example to help explain how predicted win percentage affects your SR. As an example, let’s say you play 10 games in a row and all 10 games only have a predicted 40% win rate. If you win 4 of those 10 matches, your SR be the same as it started.

Now, let’s look at all teams consisting of all solo players. They play against other all-solo teams 73% of the time, with a 50% win rate. Meanwhile, they’ll very seldomly be matched against a full six-stack (only once in every 1150 games) and they’ll have a 41% win rate. That percentage isn’t great, but it doesn’t happen often, and 41% is NOT unwinnable. If you then combine all the separate chances to encounter all the other possible enemy team compositions and their win rates, then the solo-only team composition has a total observed win rate of 49.5%.

However, when you decide to queue for a competitive game as a solo player, we don’t only place you in all-solo compositions. Sometimes, you’re in a 2,1,1,1,1 composition which has an overall observed win rate of 50.03%. Other times, you’re in a 2,2,1,1 composition with a 49.46% win rate, or a 3,2,1 comp with a 49.93% win rate. All of the possible combinations considered, the combined win rate for solo players end ups being 49.94%. That’s very close to the perfectly fair value of 50%, which means that having groups and solo players queuing together has nearly no effect on a solo player’s win rate.

Among all possible team compositions, the one with the highest win rate is the full team of six players. If we again take into account how often the six-player team composition plays other compositions (and close to 84% are against teams with at least a four-, five-, or six-player group) then they only have an observed win rate of 52.88%. So, there is a very modest advantage to playing in a six-stack. All the other composition possibilities have closer to 50% win rates.

That’s a lot of math and data to demonstrate that playing solo or playing with a group doesn’t have much of a global systemic effect on your win rate or SR! That being the case, let’s talk about how grouping DOES provide a host of advantages that definitely make playing Overwatch more FUN.

The greatest benefit of grouping is the simplest: You get to play together with your friends! There’s really no better way to play Overwatch. You can laugh at each other’s jokes and silly plays, celebrate together when someone does something awesome, and just enjoy hanging out while pushing a payload or defending a control point. Losing doesn’t feel so bad and winning feels even better.

With the new Looking for Group tool, there are now even more opportunities to play as a group, find new friends, and have a better play experience. Whether you create or join a group, the tool allows you to find other players who share your goals. We also sort the groups in the LFG tool by SR and connection quality, so you can more easily find the best group for you. Grouping with like-minded players gives you a great head start to a more positive experience.

  • Grouped players are often more willing to positively communicate with each other.
  • Grouped players are better able to resolve conflicts concerning player roles and team compositions.
  • When part of a group, you’re better able to handle adversity. All teams lose fights and experience setbacks during a game. Grouping helps make players more resistant to catastrophic tilt and go on to recover and win fights and games in the future.
  • If you find new people that made your Overwatch matches more fun, you can add them to your friends list and play with them again.

In spite of all this, some players think that that grouping isn’t a good idea. They’re afraid of being matched against other groups who may have been playing together for longer, and thus have greater synergy. There are several issues with this line of reasoning.

  • Most groups don’t play very long together with the exact same lineup. It’s actually pretty rare for the same group of six players to play together for a long time. It’ll likely become even rarer with the Looking for Group feature, which makes it easier to create groups.
  • Many of the benefits of grouping manifest soon after players come together. These benefits can increase the longer a group plays together, but they have diminishing returns.
  • If two groups have similar mechanical skill and individual decision making, but one of the groups has a formed a synergy advantage from playing together for a long time, that advantage will naturally result in a higher SR. That means those two groups are not likely to be matched together because their players will have different Skill Ratings.

Another reason players don’t group up is that it can be difficult to form groups. This is precisely the issue that the new Looking for Group tool addresses. If you enjoy playing tanks, you can find groups who really need you to hold the line. If you want to play with a very specific team composition, you can create a group that mandates it. The Looking for Group tool provides a lot of great flexibility, allowing you to both create your own or join other players’ groups.

OK, that was a lot of words. For those who have read this far, I hope it was all helpful and educational. In conclusion, I’ll end with some bits of wisdom:

  • Don’t worry too much about SR. Have fun, play to learn and improve, and your SR will rise over time.
  • Playing games with other people is pretty awesome. The LFG tool helps make this happen, so give it a try.
  • Be good to one another. Life’s too short.

Thanks, everyone!


QuickPlay (Fix It)
How Competitive Skill Rating Works (Season 10)
What's the New Deal with MMR?
Will skill rating decay at 7 days or after?
How Competitive Skill Rating Works (Season 10)
Grouping in comp inefficient? (not LFG related)
Why won't Blizzard reveal how matchmaking works?
Why won't Blizzard reveal how matchmaking works?
I, a top 500 player, am quitting overwatch, or going to barely play. Here is why Blizzard needs to wake up
Hiding profile was a bad move
LFG Feedback So Far
Why Handicapping (MMR) is Wrong for Competitive Play
GUIDE: How to use the Looking for Group System
Looking for Fair Games... Really?
What's the New Deal with MMR?
Questiom about placement matches
Toxicity is worse than ever
People are leaving overwatch because bastion needs a nerf
Blizzard encourages solo queue?
How Competitive Skill Rating Works (Season 11)
Thinking of returning to comp after several seasons - will MMR reset?
(Tom Powers) #2

#3

Wow, thanks for posting! This answers a LOT of questions!

TL:DR for Scott’s Post

Introduction

  • The new Looking-for-Group feature is coming soon to Overwatch.
  • Many players have misconceptions about grouping up together.

About Skill Rating Adjustments

  • Your SR adjustments are not affected if you are playing in a group or playing solo queue.
  • Your SR adjustments is purely based on your individual skill rating not the average skill rating of your team or group.
  • You gain more on a win if your predicted win chance was <50%. You gain less on a win if your predicted win chance was >50%.
  • Players who play less consistantly will have increased skill rating adjustments.
  • Players who decayed in Diamond and higher will earn more SR until their hidden MMR and skill rating equalize.
  • Players in Platinum and lower have a small influence based on their personal statistical performance in each match.
  • The higher you reach to 5000 SR, the smaller gains you get for winning matches.

Facts about grouping up.

  • The system does make every attempt to match groups of equal size as close as possible to each other to provide the most fair play experience for all players.
  • A match cannot be created if the expected win percentage is less than 40% for either team.
  • A six stack has the highest possible win percentage on average in Overwatch at 52.88%
  • The looking for Group feature will organize listings by your skill rating and by connection quality (which servers you connect to).

Benefits to grouping up

  • Playing in a group is less toxic and a more positive experience where players are willing to work together versus playing against total strangers.
  • There is a misconception that grouping up has no long-term benefit, due to the fact groups don’t stay precisely the same or don’t synergize as well as other groups.

Questions the community is asking

  • Do you believe the SR restrictions for grouping (1000 SR for Bronze through Diamond, 500 Masters, 250 Grandmasters) is still fair? There have been players calling for more restrictions in lower ranks, however, we know there has to be a reasonable balance between that and the ability for friends of similar skill to group together.
  • It is believed extremely high ranks such as Top 500 range have difficulty finding matches as a 6-stack group. Is this a misconception? If it is not, what can be done for players of any rank be able to group up and find fair matches within a reasonable amount of time?
  • Do you believe the penalties for leaving games in Competitive Play is still balanced? As one of the forum MVPs I answer dozens of questions each day in the tech support forum about players who complain about getting penalized when they disconnect. On the flip side of the coin, I answer nearly as many questions here on the Competitive Discussion forum about players who complain about leavers and disconnecting players and why remaining players are punished. The last available answer is way back when both you (click for source) and former community manager Lylirra (click for source) explained that SR adjustments are not accounted for leavers in order to minimize leaving and toxicity in games. I would love to confirm that the balance for penalizing players for leaving competitive games is where it needs to be. (I personally believe it is.)
  • From the user Moosemoose: Since we’re on the topic of SR in general, do you have anything you can share with us about players who have reported that while unable to climb on their primary account, their secondary started them at a significantly higher SR? There seems to be this idea that no SR resets are artificially holding some people back because of their match history.

I will add to the list if other common questions appear in this thread…


We all crashed. We demand our 50 sr back
#4

Well said, this well improve the quality of Overwatch


#5

Having used this a fair bit on PTR, it along with commendations helps make Overwatch feel a whole different game in competitive, along with the Symmetra changes!

I definitely recommend grouping up…I’ve been a solo queue sufferer a lot of the time, but have no reason not to now :slight_smile:


#6

Not bad… Looks like this will bring the competitive match matching alot easier.


#7

so excited! loved all of the math behind this, you’ve done a great fun job at explaining it.

as someone who avoids comp but loves the intrinsic team nature of it, i cant wait to hop into other groups and gain more experience playing at a competitive level…!


#8

looks great, the lfg is gonna raise this game to its fullest potential. Plus symm meta :smiley:


#9

Solo player here. My SR was above 2700 once, now it’s down to 1900. The current system is bad for the SR of solo supports.


#10

Ok
but who is hero 28?


#11

BEST. ADVICE. EVER.


#12

Ok, this got confusing and seems to be contradictory.

If your SR is determined by what the matchmaker determines is your chance of winning. Chance of winning increases by group size.

So hince, if you have a group… and you get matched up with a smaller group… then your chance of winning is higher, so your potential SR win is small and loss is greater. But then you get put up against another 6 stack and you are favored to win because you’ve been beating lower grouped teams… and you lose… you will end up with a net loss in SR even going 1-1.

This is exactly why ppl don’t group up. It is stastically worst SR gains because the chances of you getting put into a game where you are at a disadvantage is less then if you solo queue or with smaller groups.


#13

Thanks for the detailed explanation, I’m looking forward to when ‘looking for group’ actually goes live which hopefully isn’t much longer.


#14

there is something you kind of avoided and could be severely missinterpretated, your SR wins or loses depend on the winchance and the group vs personal SR. im 2700 and i play with my cousin at 2100, we get 2400 aprox games, several of which our team has LESS sr, yet when i win/lose i get 20 and lose 27-30, and thats with less win chance, i payed special attention to both groups SR average for this. Also as you can imagine my cousins wins/loses in SR are the oposite, this has been a constant for several seasons.

So either thats intentional and quite important, which i assume because it makes a lot of sense especially to avoid exploits, or there is a bug and the win prediction isnt the only factor, yes im aware of performance and i had some reaaally good matches when losing and still got the same numbers, my point is, the difference between your SR and group average is VERY significant.

i did an excel to record all of this which i dont have at hand right now but this arent just some random numbers and bias.


#15

Well, you know… when this is a promotional piece to try and get ppl to group up and use the LFG system… you gotta exclude some things from tbe equation… otherwise people will still not group up.

It sounds like they haven’t done anything to change the match maker to incentivize grouping up.

This is exactly what everyone who was against a LFG feature said as to the reason a LFG feature wouldn’t work.

The problem is people don’t want to group Scott, because of exactly what you said on your post.

What you consider fair matchmaking is exactly what makes it difficult to get a fair SR gain… because the deck is stacked against you.


#16

I have a question:

Does a team with 6 support mains have the same calculated win percentage as a team with 2 supports, 2 tanks, and 2 damage? If SR is the only thing that is taken into account when calculating a team’s chance of winning, how can it say it’s a fair match when people are forced to fill a role they aren’t going to play their best in?


#17

I think can answer this, it has been confirmed the hidden MMR and Skill Rating does not take into account the playtime of your heroes. This is why it is important that you are trained in different roles or group up in advance when you play.


#19

Scott, this is fantastic information.

I’d like to add one more positive element to grouping, especially with a regular group of people.

Mistakes. When I make a mistake, and I certainly do, that regular group makes it easier for:

  • Me to admit to the mistake.
  • My group not to get salty with me.
  • Advice to be given and received in a positive manner.

Grouping to put it simply, provides the best environment for growth and improvement in any co-operative game.


#20

True story. Where’s the fun in getting angry? You’re bound to have good and… not so good matches!
Bottom line: This game is crazy fun :smiley:


#21

Exactly… he didn’t answer the most important question. What does into determining what the computer determines is an advantage.