How Competitive Skill Rating Works - Season 17

Kaedi-11739 · July 1, 2019, 10:11pm

SEASON 17

By Kaawumba • Season 17 Edits by WyomingMyst

Welcome Overwatch competitive players. This is the Season 17 guide to How Competitive Skill Rating Works. This season I, WyomingMyst, have taken over as Kaawumba is moving on to other ventures outside of Overwatch and I wish him luck. For the most part, all information will be kept the same except for any new updates that I have documented. I have made a few minor tweaks to improve readability and organization in this thread. Remember this guide helps explain possible scenarios to skill rating adjustments that you may encounter as you play in Competitive Mode. This guide does NOT provide any relevant information to game or rig the Competitive System (that is because there is none). Furthermore, I, WyomingMyst, do not represent Blizzard, and all information contained within is based on information provided publicly by the developers or by research from the community.

Changes in Season 17

Revealed in late season 16, during patch 1.37.0.1, any matches that have been proven to have a cheater that is banned from Overwatch will now be nullified from the records with no players on either team receiving a win or loss and all statistics discarded from the match. (It is not known if Competitive Points earned from winning or drawing a nullified match are retained.)⁶⁵

UPDATE: Season 17 will be ending earlier than the normal two-month season. It will end August 13th (exact end time is unknown). This is due to significant upcoming changes involving not only Competitive Play but Overwatch overall with the upcoming 2-2-2 Role Queue. To learn more about these changes that are currently available for testing on the PTR, please see the official Overwatch news post.

Kaawumba's Introduction

The rating and matchmaking system is confusing, and a good overview does not appear to be available. The official overview^{1, 48} is incomplete and does not answer a number of common player questions and concerns. This information below is gathered from sporadic blue posts and developer update videos, and salted with my own experience and experiments, various forum threads, and watching streams. Note that since Blizzard does not give exact algorithms, I do have to fill in some gaps, or leave some items unknown. Finally, Blizzard has developed a habit of making changes without notice, so just because something was true a few months ago does not guarantee that it is still true.

Summary

See the later sections for a more in-depth exposition, as well as references to supporting material.

How do I maximize my Skill Rating?

In order:

Win games
If diamond or above, play a minimum of 4.67 games per week (starting one week after placements).
If platinum or below, perform well, statistics-wise, with the chosen hero.

In my last game, my Skill Rating went up/down by X. Why did this happen?

The details of Skill Rating (SR) movement in a match depends on whether a player wins or loses, the Matchmaking-rating (MMR) “strength of schedule” between the two teams, the variance between the player’s SR and MMR, whether a player is new, whether a player is active, the player’s statistical performance, the player’s position on the ladder, and possible bugs. Add it all together and it is nearly impossible to determine why a player’s SR has a specific movement in a particular match. Sometimes over the course of twenty to one hundred matches it can be seen that one particular cause is dominant, but that often depends on careful data collection and some guesswork.

Summarize matchmaking, rating, and progression for me

MMR is a hidden number⁵⁸ that goes up when you win, and down when you lose. How much it goes up and down is a complicated formula, based on many factors, but the essential truth of wins are good and losses are bad is certain⁴.

That MMR is then used to match people with and against people of similar MMR, in an attempt to create a 50% match³². Here is the key part:

If a player’s MMR is wrong and too low, then the odds to win will be greater than 50%, and the player will win more games than he loses, which will cause his MMR to rise over many games played. He will then be placed with stronger and stronger opponents (and stronger and stronger allies) until his MMR is correct, and his win percentage approaches 50% (with some random oscillation around 50%)⁷.

Once a player finds himself trapped in a rating range, the only way to break out is to improve as a player and play enough games to overcome any random factors^{47, 48}.

What is Skill Rating (SR) then? Skill Rating is a visible and friendlier approximation of MMR²³. It has no more meaning with respect to matchmaking than tier icons. However, MMR and SR are closely linked (except for top players who have decayed), so inspecting SR typically gives a reasonable estimate of a player’s MMR²².

Ranks

< 500 Bronze, SR not listed
500-1499 Bronze
1500-1999 Silver
2000-2499 Gold
2500-2999 Platinum
3000-3499 Diamond
3500-3999 Master
4000-5000 Grandmaster
Top 500 players in region.

For season 3, the rank distribution was Bronze: 6%, Silver: 22%, Gold: 34%, Platinum: 23%, Diamond: 10%, Master: 3%, Grandmaster: <1%³. For season 8, the rank distribution was Bronze: 8%, Silver: 21%, Gold: 32%, Platinum: 25%, Diamond: 10%, Master: 3%, Grandmaster: 1%²². Third-party sites such as masteroverwatch.com and overbuff.com do not give reliable distributions because players have to actually log in to those sites for it to count the data and profiles that are listed as private alter those statistics even more. This skews those sites heavily toward higher ranked players.

As a player gains SR, he will be promoted to the next tier when appropriate. He will drop out of each tier as he loses rank. However, there is some loss protection for Diamond and below. After a loss (but not a win), the game will check to see what a player’s current tier and skill rating are. If his skill rating has not exceeded the minimum for his current tier for the past five games, he will be demoted ¹⁶. For Master and Grandmaster, a player will be demoted immediately if his skill rating is not high enough. Competitive point rewards are based on the highest rank achieved during the season¹⁶. Ranks and SR are wiped each season (but not MMR, see below).

The top 500 leaderboard becomes available two weeks after season start. To be eligible, a player must have played at least 50 games that season⁴⁵, all from one region. The player must have had SMS protect enabled for all 50 of these games ^{50, 51, 53}. The top 500 spray and icon are not awarded unless a player is in top 500 at the end of the season¹⁸.

Matchmaking

Matchmaking is based on hidden match-making-rating (MMR), not skill-rating (SR)^{3, 21, 25, 40, 59}. Competitive MMR is separate from the MMR of other modes³⁷.

The game’s matchmaker uses MMR to determine the numerical approximation of each player, and attempts to match teams such that the game is statistically fair (each team has a 50% chance of winning) ³⁷. Upon victory, the winners receive an increase in skill rating, and the losers receive a decrease in skill rating. At times (off peak-hours or very high/low rated players), finding a fair match will be impossible in a reasonable amount of time, and the match maker will match teams based on a best possible fit. However, note that if no matches can be made with an expected win percentage between 40% and 60% (the infamous Brazilian six stack playing at 3 AM), then no match will be made and your search queue will continue until there is an acceptable match⁴⁸. In the middle ground, top (or bottom) players will have a longer, but not infinite queue, to attempt to find a better match¹⁵.

If the match making algorithm determined before the game that it was not fair, the higher ranked team will receive less SR/MMR on a win, and lose more SR/MMR on a loss. The lower ranked team will receive more SR/MMR on a win, and lose less SR/MMR on a loss (2). This is standard for Elo (rating system from chess) type systems. The SR/MMR penalty/bonus is designed to be fair, such that if you win 40% of your games with a predicted 40% expected win probability, your SR/MMR does not change (48). Draws cause no change in SR and have an unknown effect on MMR.

Matchmaking and groups

In addition to MMR, the match maker attempts to match on ping and group size^{32, 48} and to make matches reasonably quickly. (Jeff also mentions quarantining new players from the general population, but this effect does not apply to competitive).

Predicted win probability (which is used to calculate fair matches and MMR/SR rewards) does not depend on grouping⁴⁹. This leads to a small handicap for solo players vs groups, and smaller groups vs larger groups, for when the matchmaker cannot find groups of equal sizes to match against each other. The actual (as opposed to predicted) average win rate for solo players is 49.94% and for six stacks is 52.88%⁴⁸. Smaller groups assumedly have win rates greater than 49.94% and less than 52.88%.

However, if a player groups with someone with whom they have poor synergy, it acts as a penalty on their practical win likelihood, because they are likely to be matched against groups of equal size, with more synergy and experience. Finally, large groups tend to have longer queue times while the matchmaker tries to find other large groups. With the LFG tool now being live for competitive, this may be less of a problem as there are more groups in the system.

If the matchmaker says most games are fair, then why are there so many stomps?

There are many reasons:

Overwatch, as a game, has a tendency to snowball. The winner of the first fight has an ultimate advantage that has a tendency to last the round. Losing teams tend to tilt and start playing poorly, which can continue in the same round and carry into the next round.
Random variables are random. Maybe a cat walked on a keyboard or someone fat fingered an ult⁷. Maybe someone who has 99% up-time for their internet had their 1% failure that day. Maybe one team has all dps mains, and the other team is well balanced. MMR and predicted win percentage only has validity over many matches, not each individual match.
Not every player tries hard every match. Sometimes this is subtle, like playing with a beer or two too many. Sometimes it is less subtle, like practicing a hero a player is not good at, regardless of team comp or map.
Some players actively break the system, by either hard throwing, that is intentionally losing (as opposed to soft throwing as in item 3), playing on someone else’s account to boost it, or other similar actions. These actions are bannable and should be reported when seen. Blizzard has promised to take stronger actions against this sort of thing¹⁷. However, soft throwers and hard throwers can be difficult to distinguish, and Blizzard has to error on the side of caution to avoid banning the wrong people, so it will always be a problem. The in-game report UI, as well as a key blue post³⁶, give guidelines on what behavior is or is not bannable.
New accounts in general have less data and will have a less accurate MMR. There is no required minimum number of games to play, so new accounts can stay “new” for a very long time.
MMR itself is only approximate. See the section on “How Accurate is SR”, below and realize that MMR generally will have the same issues as SR, with respect to accuracy, except it is harder to measure what is going on with MMR.

It is possible to quantify many of these effects using simulations⁵⁷.

Many of these effects are removed or mitigated by the Looking For Group feature (or building your own team by some other means), so I recommend using Looking For Group if you are in a rank/region/platform/time/day with enough players for it to be viable.

Okay, stomps are going to happen, but why are there so many win/loss streaks and large SR/MMR movements? Shouldn’t a player’s SR/MMR be fairly stable once he gets to the proper rank?

Win probability changes slowly with rank because there are so many random factors in each individual match. Unfortunately, it follows from this that frequent and long streaks will occur, and a player’s rank will oscillate widely. Essentially, a player will tend to bounce between the range of where he is nearly guaranteed to win and where he is nearly guaranteed to lose. The range varies from player to player, but +/- 250 SR/MMR is common and +/- 500 is possible. This problem can be analyzed in depth, mathematically³⁴.

Skill Rating

Skill Rating (SR) should be thought of as a visible and friendlier approximation of MMR, similar to tier icons. With the exception of Diamond and higher ranked players who have decayed and the disconnection penalty (see below), MMR and SR are closely linked²². The SR/MMR convergence is implemented by having each match pull SR in the direction of MMR. That is, if SR is lower than MMR, a player will win more SR than MMR on a win, and lose less SR than MMR on a loss. When I refer to an SR buff, this is how the buff is implemented¹⁴.

Typical SR movement has range of 20-30. Neutral SR movement (where the teams are balanced, there is no performance SR, etc.) is 24⁵⁴. Typical MMR movement is difficult to quantify as it is invisible, but is likely similar. If an account is inactive for a while, after the player returns SR/MMR gains will be exaggerated again for a time⁴⁸. For this reason, players that only do placements each season will tend to see high volatility in their rating each season.

In addition to Elo type effects for high rated players (in which they gain less SR/MMR because there are no fair matches available), there is an extra SR debuff for high ranked players to prevent them from getting to the SR limit of 5000. The concern here is that players would start to pile up at the limit of 5000 and break the leaderboard^{35, 48}. This debuff starts at 4200⁵⁴ and gets progressively stronger as the player ranks higher.

New accounts

If a player has never played competitive, his initial competitive SR is around 2350³⁸. For new players, MMR and SR are more volatile³⁵ and SR has been measured (for average players) to approximately follow the formula: SR change = -3.85 * game + 99.34³⁸ until about game 18. Top players gain more than this, and have increased volatility for 25 games or so³⁸. Bottom players assumably gain less, but no one has tested and shared the results. The maximum possible range of placements is roughly high bronze to low master, but most accounts tend to be artificially close to high gold. In addition, because initial placements depend on only 10-25 games, initial placements have substantial random causes and are frequently inaccurate.

If MMR are SR are so similar, why have both?

Let’s summarize the differences between MMR and SR:

MMR:

Not human readable⁵⁸
Blizzard’s best estimate of your skill²³
Only affected by what happens in a match⁴
Used in matchmaking²³

SR:

Human readable
Chases MMR (SR gains more on a win / loses less on a defeat for MMR > SR)²³
Affected by what happens in a match, decay, and DCs^{23, 54}
Not used in matchmaking^{3, 21, 25, 40, 59}

The real purpose of MMR is to provide Blizzard’s absolute best estimate of each player’s ability, while SR is to give a friendly indication of rank to players, as well as penalizing for decay and DCs (without affecting matchmaking).

However, it is probably worth coming up with a simpler system to prevent much of the confusion that circulates around these terms.

But I just read this post from a developer, and it said matchmaking is based on SR, not MMR

I’ve already referenced all the posts indicating that indicate that matchmaking is done by MMR alone ^{3, 21, 25, 40, 59}. However, there are also four posts that indicate that matchmaking is either based on SR or matchmaking is limited to a given SR range ^{5, 7, 35, 48}. Scott Mercer is the author of two of the yes SR posts, as well as two of the no SR posts, so he contradicts himself. The contradiction must be resolved somehow, and I decide in favor of SR being used in no way during matchmaking for the following reasons:

Whenever MMR and SR are discussed in the same post (including by Scott), it is to say that SR is not used in matchmaking.
Master+ players regularly see decayed “diamond” players in their matches. And when someone like Seagull decays down to diamond on stream, he is still placed in grandmaster / top 500 matches. If SR was either used for matching, or the limits on matching were restricted by SR, this should not happen.
SR and MMR are closely linked ²², except for decayed players (who only exist in diamond+) so saying matching is done on one or the other is a distinction without much of a difference. This makes Scott being sloppy with the terms not as surprising and makes the answer to this question less important.
There hasn’t been any clear revoking of the old system on this point so the belief that we should only keep the newest non-contradictory statements ^{35, 48} is questionable.
Generally the people who are insistent that SR and MMR are both used in matchmaking are using it to infer some broken or rigged system with horrible and implausible outcomes that are contradicted by careful analysis, common sense, and unrelated blue posts.

An alternate (but acceptable) hypothesis, in which I completely believe new posts (and declare obsolete inconvenient old posts) would be the following: MMR does not exist anymore. Posts describing MMR in competitive are obsolete. Match making is based on SR alone ⁴⁸, except in the case of decayed players, where there is a specific hack in the code to use their “internal skill rating” instead of their “displayed skill rating” in match making. I’m not ready to change my whole post to support this hypothesis, as it hasn’t been explicitly endorsed by Blizzard, though it would be the correct conclusion if you ignore posts from February 2018 or earlier.

What would not be a reasonable conclusion is that matchmaking is based on some amalgamation of SR and MMR that ends up rigging matches or breaking the system. Regardless of what we call the number used to make matches, it goes up when you win, down when you lose, and is a single number.

I have tried to ask Scott for clarity on this issue, but have not gotten through the noise.

Rank Decay

To prevent rank camping, for players ranked 3000 SR or higher, their rank will decay 25 SR per day if they do not play. Each game a player plays increases his buffer by 36 hours, to a maximum of one week²⁰. Each day a player’s decay buffer decreases by 24 hours. If it hits zero, decay starts. To determine how many games must be played per week to avoid decay, we can calculate (1 game / 36 hours) * (168 hours / week) = 4.67 games / week. This is slightly lower than the 5 games per week that was originally reported¹⁵. To see if decay is imminent look on the right hand side of the information screen of the competitive play card. After returning from decay, the player will have a substantial SR buff (gain more SR from wins than he loses from loses) until he is back where he was^{3, 48}. While decayed, a player’s MMR (and hence matchmaking) does not change ³. The decay clock doesn’t start until the player does his placement matches.

Decay can be quantified⁵⁴. In brief, at extreme decay, more than 100 SR is gained per win, and 2 SR is lost per defeat. For slight decay, the SR gains are modified by only a few points. Recovery from decay is a linear relationship between change in SR and SR - hidden SR (what your SR would be if it had not decayed, not MMR)⁵⁴. The motion in SR is larger the further hidden SR and SR are from each other.

Performance Modifier

In platinum and below, SR/MMR gains are adjusted up or down based on the performance of the player. This is a minor factor^{2, 48}. This is done based on a numerical comparison of measureable quantities such as elims, deaths, assists, damage blocks, ults cast, etc. between a given player and other players of that hero at that MMR⁸. Generally it is assumed that the measured stats are those visible to the player, but that has never been confirmed by Blizzard. Most of the details of this implementation are fuzzy and not published (probably to reduce exploits). This performance measure is correlated but not identical to “on fire” calculations. “On fire” compares a player to his teammates, while SR/MMR bonuses compare a player to other players that are not in the current match, but in a similar skill bracket, and playing the same hero⁸. The effect of the performance modifier is generally small, but there are plausible reports of it causing people to have to maintain a 55%+ win rate to maintain their SR¹¹.

This is the most controversial (among many controversies) part of the SR system. There are two camps, those who don’t want their SR/MMR to be so heavily influenced by those in their group, and those who worry that having a performance based system will cause people to not play the objectives / win conditions and instead go stat hunting. In addition to people who go stat hunting, people can unintentionally be at the wrong rank because their good / bad play is not reflected in their stats. In the early days, these groups were roughly equal, but lately those advocating for no performance modifier are more numerous.

Starting in season 8 (January 2018), the performance modifier was removed for diamond and above³⁵. Blizzard has started to realize that having a motivation other than winning causes all sorts of non-ideal behavior and effects. The immediate forum response has been to request this change for lower tiers as well (at least whatever tier the poster happens to be in). However performance modifiers do help move new players where they belong much faster than an Elo type system, so I propose that the performance modifier be removed for Bronze to Platinum players after a sufficient number of competitive games has been played. Blizzard can use their data to determine how long it takes for people for their rank to stabilize, which would be a good place for a cutoff.

Season Transitions

Players’ 10 placement matches will start with Blizzard’s best estimate of each player’s MMR for matchmaking purposes¹⁵. A player’s initial SR will generally be close to his previous season ending SR, with a relatively ordinary adjustment based on 10 placement matches.

However, occasionally a player will have a large movement²⁴ in either direction. This happens on relatively inactive accounts³⁹, and can be explained by Scott’s statement: “Play a lot of games, it (MMR) gets more certain. Don’t play Overwatch for a while, it gets less certain. The more certain the matchmaker is about your MMR, the less your MMR will change in either direction based on a win or loss”^{2, 48}. Essentially, if a player only does his placements, or slightly more than that, in a season, his MMR/SR is prone to much wider swings when he returns.

Because of reasonably high quality data for seasons 6, 7, 8, and 9²⁴, it is possible to derive a formula: (New Season Starting SR) = (-173.7 +/- 20.8) + (1.0015 +/- .0059) * (Previous season Ending SR) + (37.94 +/- 2.23) * (Placement Wins). New Season Starting SR has a standard error of 181 SR (24), which implies a possible change in rating during placements of +/- 500 SR for inactive accounts. For active accounts, the max range is more like +/- 200 (from winning or losing all matches).

Decay does not persist through season transitions. That is, if a player decays from 4000 to 3000 before the season break, he will typically place around 4000 after the season break. Of course, if he is an inactive player (which decayed players often are), he may see an unexpectedly large swing in either direction.

Prior to season 6 (September 2017), players’ SR would be bumped down below their MMR at the beginning of the season, and they would earn it back with an SR buff over the course of 50 matches. This felt bad and was removed¹⁵.

Win Streak Bonus / Loss Streak Penalty

Prior to season 5 (June 1st, 2017), there was a substantial bonus to SR for winning many games in a row (about 4 or more), scaling all the way up to 150 SR for one win. To keep things balanced, penalties existed if a player lost many games in a row. This was removed because it lead to people who got lucky/unlucky being thrown far from their true rank^{9, 10}.

Sometime since then (the first rumor I have is from season 11 (July 2nd, 2018)), streak bonuses returned, in a more limited form. This design is much less likely to throw people out of their rank incorrectly, while still allowing accelerated motion for people that really don’t belong there.

My inferences for the rules are:

If a player has very high statistical performance AND
A high win percentage AND
Is on a win streak AND
Is below diamond
He gets highly exaggerated gains.

The reverse also seems to be true:

If a player has very low statistical performance AND
A low win percentage AND
Is on a loss streak AND
Is below diamond
He gets highly exaggerated losses.

The maximum gain/loss (including performance and streak bonuses) seems to be a little over 100.

But because of all the conditions, and the relative lack of data, it is difficult to be sure. I especially don’t know if high win percentage or high statistical performance or both is key. My data is the following: Win streak bonuses are clear in two recent bronze to GM streams⁶⁰, which occurred below diamond. There was no such effect in diamond+ accounts⁶¹. I have an anecdote that losing streaks (on an established account) below diamond behave similarly⁶². Finally, a bronze player, with bronze statistics, won six in a row with no noticeable bonus⁶³.

How Accurate is SR?

On the forums there are generally two factions with respect to SR, those who think that SR is essentially luck and a meaningless value, and those who believe that SR is essentially correct. Neither of these factions have the correct approach. The correct question is to ask how accurate is SR? Clearly top 500 players are better than bottom 500 (see any bronze to GM series or watch low bronze play vs top play), so SR is not completely random, but how accurate is it really? For this discussion, I am assuming that we are talking about a player that plays to win every game, doesn’t share his account, and has played at least 100 competitive games, and continues to be an active player.

There are number of ways to approach this question. One is to start a completely new account, and then play 100 games on the new account and see how it performs compared to the old one. This shows that SR can vary by 1000 SR in extreme cases, and 500 in normal cases²⁷. There is some evidence that reroll experiments show less variance at higher ranks²⁸ likely because there are less random variables, such as smurfs, throwers, and inconsistent play.

Next, any player can see how his SR changes during a season. A range of 500 is completely normal here.

In addition, if an account needs a 55% win rate to maintain SR¹¹, and if win % changes slowly with rank, then it is expected that this will be an additional source of significant error in SR.

If a player gets noticeably better, it can take a long time to get to the correct rank. If a player maintains a 55% win rate, he will only go up approximately 220 SR/MMR every 100 games. Because there are twelve players in a match, one player’s contribution (and ability to carry the match) is limited unless he is playing at a vastly different skill level, so a 55% win rate for a player that is moderately under placed is to be expected.

Put this all together, and we can state that an active and motivated player’s SR is only accurate to +/- 250 SR in normal circumstances and +/- 500 SR in extreme cases. Of course, new players, barely active players, or players that actively break the system can be off by much more.

MMR/SR Reset

Periodically, there are calls on the forums for all MMR/SR information to be wiped and to start over. The justification here is that because the system has had many flaws, and that MMR/SR takes a very large number of games to move, and that people are misplaced by 250 to 500 SR (see previous section), that it would be good to wipe everything and start over. The chief problem with this is that before the wipe, the typical error is +/- 250 SR, with +/- 500 SR in extreme cases. Immediately after the wipe, the typical error would be +/- 2500, as everyone would have the same rating. Top 500 would be playing against bottom 500. This would cause matchmaking and rating quality to fall through the floor until a sufficient number of games are played. And because many people do not play a huge number of games, or play inconsistently, this inaccuracy would persist for months. And at the end of that, the accuracy would be similar to what it was before the reset, because the underlying system has not changed. My reasoning against a reset is similar to that given by Scott¹².

Popular Myths

There is a lot of repetition in these subsections. My apologies, but the various myths tend to have a lot of overlap. They also have incompatibilities, so it is difficult to address them except individually, with repetition.

Forced loss streaks

There is an old and persistent conspiracy theory that Blizzard’s algorithms force players to have a 50% win rate by nefarious means (if a player is doing well, very poor players are put on his team to make him lose, often for many games in a row). This has been contradicted by Blizzard^{7, 46}, is contradicted by people’s ability to climb¹³, and would be a horrible and difficult-to-implement design. The truth is much simpler. If a player wins more than he loses, his SR/MMR goes up. As it rises, he is placed against stronger opponents (and with stronger allies), which increases the chance that he will lose⁷. Once in equilibrium, the average person he faces (and is allied with) is at his skill level, and the only way to go up is to become a better player. The win/loss patterns and streaks are fully consistent (in the mathematical sense) with the system as Blizzard describes it, and are not consistent with a system in which Blizzard forces wins and losses³⁴.

Matchmaking takes into account win percentage

It explicitly does not. “At no point in MMR calculations do we look at your win/loss ratio and win/loss ratio is never used to determine who to match you with or against. We are not trying to drive your win/loss percentage toward a certain number” – Jeff Kaplan (32). However, probability of winning depends the difference between MMR and a player’s skill level, and MMR does depend on how much a player has won or lost lately, so in this way there is an indirect dependence of win probability on win/loss history. This leads to (correct) statements that “Whenever I reach, or am near, my career high, I start losing and fall back down”. This corresponds to a player reaching his skill cap. The only way to break out of an MMR/SR range is to improve as a player and play a sufficient number of games.

Matchmaking takes people at a given SR, and then matches high skill players with low skill players to make a balanced match. Sometimes this is called handicapping, similar to amateur bowling or golf. This is done to achieve a 50% win percentage

This is not compatible with Blizzard’s statements: “All the system does when it comes to matching on skill is attempt to match you with people of a similar number”³², and “We use MMR for matchmaking, not SR”²⁵ and “SR closely chases your MMR up and down and is a more “digestible” number. With the exception of top players who have decayed, MMR and SR are closely linked”²². It is also contradicted by people’s ability to climb¹³.

MMR is determined by statistical performance

The idea here is that a player’s MMR is really a summary of their statistical performance, and that because SR chases MMR¹⁴, a player’s SR is basically determined by their statistical performance. If this were true, statistical performance would be, by far, the most important contribution to SR. This is contradicted by Blizzard² and by most players having a win percentage of about 50% (if statistical performance was weighted very heavily, stat-chasers would have much lower win percentage at a given SR than team-players, for example).

Matchmaking pushes a 50% win percentage using other broken criteria

It varies what this broken criteria is. The theories typically have a few things in common:

Blizzard has never confirmed (and sometimes has denied) them.
They would lead to obviously broken matchmaking.
Their proponents never have solid data backing them up.
They usually are not compatible with Blizzard statements, “All the system does when it comes to matching on skill is attempt to match you with people of a similar number”³², and “We use MMR for matchmaking, not SR”²⁵.

The supposed support for these theories are that Blizzard has indicated that they desire and are happy with a 50% win percentage (32). However, there are many ways of accomplishing a 50% win percentage, and much better ways than the theories these forum writers ascribe to. Specifically, a 50% win percentage can be accomplished using a procedure similar to that which Blizzard describes (7): As a player wins matches, he is placed with and against stronger players. As he loses matches, he is placed with and against weaker players. With time, his win rate will converge to 50%, with some random oscillation around 50%. If he gets better (compared to the rest of the community), his win percentage will go up a bit until he finds his new level.

There is an SR penalty when grouping

This has been explicitly contradicted by Scott⁴⁸. If you take into account the entire system, there is a small SR advantage when grouping. This is because grouping increases your average win probability slightly as larger groups are unable to find equally sized groups and get matched with smaller groups. The system does not try to correct for this⁴⁹.

Reputation is taken into account in matchmaking

Blizzard is a company, and companies are inherently deceptive. See cigarette companies’ studies that indicate that cigarettes are good for you. Blue posts are not a reliable indicator of how the system works. My experience is proof that the system is malicious.

Yeah, well, you know, that’s just, like, your opinion, man.

More seriously, though, I take it as given that Blizzard employees don’t lie or intentionally mislead their customers. If you don’t agree, we don’t have much to discuss here. I can prove some things with data, but much of the system will always be opaque if we disregard all the blue posts. Even if Blizzard open sourced their matchmaking and ranking code, people could claim that the available code is not actually implemented in the game.

Leaver Penalty

Each time a player leaves competitive matches any time before the Victory/Defeat screen, he will receive a 50 SR penalty. In addition, a leaver will receive increasing automatic bans, with each leave¹⁹:

10 minutes
30 minutes
2 hours
8 hours
24 hours
Season Ban (with a season ban, no rewards including end-of-season bonus competitive points are received)

Recovery from disconnecting is identical to recovery from decay⁵⁴. That is, after a single disconnect a player will have an SR buff for about 12 games until his SR has caught back up to what it would have been if he’d never disconnected. Because leaver recovery is identical to decay recovery, it is likely that MMR is unaffected by leaving, just as it is unaffected by decay.

Season bans cannot be reversed. The reason or method of leaving is irrelevant⁴³. If a player plays many games without leaving, the leaving penalty resets downward. The rule of thumb is that if the bans are getting into 8 or 24 hours, a season ban is imminent. If a player receives three season bans, a permanent ban is likely³⁰. The season bans do not need to be consecutive to count toward a permanent ban.

If the player leaves before the game is 30 seconds old, the match is cancelled and no one except the leaver is penalized/rewarded. If a player leaves after the game is 30 seconds old, Players on the leaver’s team get no consideration due to the leaver. That is, they (and the other team) gain and lose SR/MMR as normal. This is to prevent situations where it is the team’s interest for one person to leave and save everyone else from losing SR/MMR (and the winners from gaining SR/MMR)^{41, 42}.

If a player leaves and rejoins the match (after rebooting / internet comes back, etc.), he will often be able to save himself from getting the penalty, but it is not a sure thing. He needs to rejoin quickly, and not disconnect and reconnect several times⁴³.

Leaving during or after the Victory/Defeat screen is safe, but it does not save time before joining the next match. The matchmaker waits for the previous match to be complete before attempting to place players in a new match⁴⁴.

Exiting the game while in a queue or a group is not safe. The game does not immediately remove you from queue and group when you exit, so a queue can pop (and a group leader can queue you up) before the game figures out you are gone and you can be penalized as a leaver. Always leave groups and queues and wait a minute before exiting the game.

NEW - Impacts from Cheating

Discovered by the community starting in late season 16 during patch 1.37.0.1, it has been discovered a match will be nullified if a player is detected to have been cheating including but not limited to the use of unauthorized third-party software. This means any player who participated in that match will have any SR adjustment reversed and any stats no longer reflected in their career profile.⁶⁵ There is no information if any Competitive Points won for that match is also removed.

Bugs

Because of the complexity of the system, subtle bugs with skill-rating and match-making can be difficult to spot and reproduce by the players. And many player reports are not particularly trustworthy for various reasons. However, there are some outstanding issues which are well-known and verified.

UPDATE: Since the start of season 17, there have been noticeable changes in matchmaking allowing faster queue times. This however has resulted in some noticeable bugs. These bugs are not yet confirmed by Blizzard but many are expected to be resolved when the new patch launches on August 13th. The first includes getting penalized if you attempt to cancel a search queue for a Competitive match. To work around this, simply avoid canceling your queue and only search when you are absolutely ready to play. The second is when a match is found, not enough players will be present. Unlike the description below recommending to leave right away, do NOT do so as this will result in a penalty. The match will cancel on its own within 60 seconds. Finally, players in extremely low bronze rank (approximately under 1000 SR) are having infinitely long search queues and are unable to find a match.

There are a number of reports of a bug that will cause people to have very small gains / large losses in the match after the match with a leaver²⁶, assuming the bugged player left after the game said it was safe to do so, but before the end of the match. To avoid this, never leave a match, even after the game says it is safe to do so. There are also a number of people who a have reported small gains or large losses without any leavers in the vicinity, so we do not entirely understand what is going on. Blizzard has said that there is a fix in the works³³ but it has been a year since that post and there are still several complaints about this issue each season. If you do encounter this problem try to provide as many details as possible in the Bug Report forum.

There was a rare and serious bug in which players can get incorrectly season banned and lose huge and undeserved amounts of SR. This bug has not happened recently. It appears that what happens is a competitive game gets put in a “Waiting for Players” state. And each time the game restarts with new players, everyone loses 50 SR and gets an increasing ban, up to hundreds of SR and a season ban. Blizzard has acknowledged the problem, is working on it, and has promised restoration to affected players³¹. However, restoration can be slow and painful. If you ever see “Waiting for Players” during a competitive match, you should exit by any means necessary to stop the bleeding. This bug has not been reported by any player in months so it is assumed this bug has been resolved.

Using statistical measures to rate players is particularly prone to subtle bugs¹¹, which Blizzard and players have been discovering and Blizzard has been (slowly) acknowledging, fixing, and re-implementing since launch.

References

Last edited on 7/27/2019 by WyomingMyst

Kaawumba-1133 · July 2, 2019, 4:11pm

Thanks again for taking this over.

Vurg-1970 · July 21, 2019, 2:39pm

Seems that there’s a bunch of good solid info from Jeff Kaplan in this interview that should clarify things about MMR and SR. Kudos to Kaawumba for the summary of the interview.

Zarathustra-21691 · July 21, 2019, 3:57pm

Nice summation, very extensive. You have some things that are contrary to each other (ie. Matchmaker tries to create ~50% winrate vs. Matchmaker will give you easier/harder matches if your SR and MMR are not aligned), but it is what it is considering how little actual info on the topic this company is gracing us with.

Kaawumba-1133 · July 21, 2019, 4:44pm

This is not true, and the article never says that.

Zarathustra-21691 · July 21, 2019, 6:22pm

Does it really not say that? Let’s reconvene. This isn’t the only conflicting information in the whole article, if you want I can list all of them? I don’t think this is yours/op-ed’s fault, it’s Blizzard’s lack of transparency.

Squirrel-12114 · July 21, 2019, 6:28pm

Because the player will make the difference as he is far more skilled than his SR would suggest, not because the game is purposely giving him rigged matches.

Zarathustra-21691 · July 21, 2019, 6:30pm

The game is ringing the matches because of the reason you stated. We can pretend all we want that it calculates MMR properly all of the time, but it turns up that it doesn’t according to the experience that most players have.

Mby nothing is wrong with the matchmaking system, but everything is with the sys to measure MMR vs SR?

Squirrel-12114 · July 21, 2019, 6:37pm

Not sure if serious. Let’s take your stats for example; if someone with say 42% sombra wep acc instead of 35, 5 elims per life instead of 2.3 and ~4 deaths per 10 played on your account at your SR they would have a far better winrate and “the system” has nothing to do with it. The only variable is you/getting better. Let’s not pretend you are where you are because of supposedly “rigged matches”.

Zarathustra-21691 · July 21, 2019, 6:46pm

Huh? Who’s talking about me? Also your grammar is a bit whack and it’s hard to understand the though experiment you are suggesting? Additionally, I have multiple accounts (this being my main) and the other placing high diamond and mid-masters … So is this MMR really correct? It might do an OK average job, but it does nothing to take into account swings in performance based on the players’ mood that day.

When the Matchmaker rigs matches for MMR vs. Sr discrepancy, there is 11 other people in the team suffering for it. Let’s say that the other 5 on your team are biased to win the game (low MMR for your real rating) and the 6 on the enemy team are biased to lose (high MMR for their real rating). Is that really a match that should be happening? Where does that fall within the ‘fair match’ philosophy? Additionally, if any of the other 11 players are not part of either groups, they are basically caught in the crossfire of the MMR system trying to fix it’s mistakes.

I wonder why everyone in diamond and above are mostly happy with their games (and don’t report 8/10 steamrolls - 4 for you, 4 for them)? It might have to do with some differences in the matchmaker between the ranks, right? What is the only difference?

Audaylon-1620 · July 21, 2019, 7:41pm

Sticky?
charcaters re

Squirrel-12114 · July 21, 2019, 7:49pm

The point is it’s the player who makes the difference especially when they are playing far below or above their SR, not a rigged matchmaker and quite frankly your stats are very plat.
I’m sorry but no mid master player would have a lvl 900 account in almost gold with such a poor accuracy and low elims per life.
The average mid master player does not believe in rigged matchmaking, elo hell or conspiracy theories, only the boosted ones do.

Kaawumba-1133 · July 21, 2019, 10:18pm

Squirrel’s interpretation is correct. In addition, note that Jeff has said once again that SR is in no way used in matchmaking, so any theory that involves looking at the difference between MMR and SR to set up matches is incorrect.

Zarathustra-21691 · July 22, 2019, 11:22am

(Ignore the following 3 paragraphs if you are interested in the topic and not some petty stat shaming)
Dude, are you reading the wording that THEY have said, and that is reinterpreted in Green by our MVP-chan here, or you are just judging me by my stats?

Let’s talk a bit about the latter. Firstly, I’ll be back in a couple of days off the mountain and will post from my other acc just so that you can bite your tongue. Secondly, you are looking at my last season stats (where I’m duoing with a friend and playing as a DPS mostly - while Diamond was reached on this acc off tanking) - but I guess we’ll take the current stats by face value and go with that. Thirdly, avg stats don’t mean squat out of 2 reasons: a)If you started as a silver player (and were there for about a year) and then clawed up do diamond - your avg stats will look like a plat-gold for the next year or so. The friend who I am playing with now currently reached plat, but his avg.stats look like a midgold (on most stat comparison websites), so I guess he should stop winning games with ham so that the stat gods are pleased. If you take a look at our Reference-Prophet’s stats here (Kaawumba) with slight-interchangability we’d have about the same avg.Total Coefficient - yet we are 300 SR apart currently. b)avg. Stats have 0 real-game translation, as you can see from streamers that stick long enough for the medals to pop-up and be compared to their career avg. Ignoring daily performance, flow, comp/team-mates luck is ignoring the reality of how this game is played. Furthermore, thinking that accuracy or high damage translates into wining games is just another step in the direction (totally ignoring gamesense/positioning and not wanting to feed ult).

It is funny (but also sad) that you got caught upon trying to discredit me more than you did arguing on the topic at hand.

Now back to it, if the player is the one making up the difference, then the matchmaker should account for that and aim for that 50% win rate anyway (by perhaps matching them against another smurf). The other 11 players suffer because one player needs to climb or drop down to a rank the internal system thinks they should be in. It is abhorrent to me how people continuously blame the community and it’s individual behavior on win/loss streaks, and do not for a second think that there might be something rotten in Sweden. I wonder why such streaks and MMR swings almost never happen in the other 3 major Esports games.

@Kawuumba, if Sr and MMR are totally unrelated, can you elaborate how the Personal SR adjustment Bellow diamond works?

Also, as noted multiple times in your collection, by yourself, there is multiple instances of conflicting information coming from different places of the OW staff. So, why, Oh why, would you believe Jeff on anything that he says(?!), when he (and the rest) have gone back and around on their words countless times on things that were this way, or that way, or were never going to happen in the game.
As noted by yourself, you usually just pick one side to believe that seems most plausible to You, so that you can get some semblance of truth and stability in your arguments. Non of this has been empirically tested with a satisfactory sample size, nor have the real specifics of the matchmaking algorithm ever been published to put an end to this debate.

Tbh, I only wanted to have a civil debate but that seems to be imposilbru on this forum as the favorite argument (see: fallacy) of choice is either ad-homimem or faulty generalizations. Henceforth, expect me to pick apart and list ALL the contrary information that are contained above when I am back in civilization.

Oldphardt-2154 · July 22, 2019, 3:19pm

These matches are absent in player profile, but SR change is there.

Sheevah-1353 · July 22, 2019, 4:55pm

For the sake of clarification, which of the following statements do you believe to be more accurate?

-The matchmaker cannot have an accurate MMR in all scenarios. When the MMR assigned to a player is inaccurate, it leads to rigged matches that may be unwinnable/unlosable for other players in the match.

-The matchmaker knows enough about players to calculate accurate win percentage chances for both sides. The matchmaker intentionally places players on sides that are more or less likely to win a match to force them to reach a desired SR number.

Kaawumba-1133 · July 22, 2019, 5:47pm

I didn’t say that SR and MMR are totally unrelated. I said that SR is not used in matchmaking. If you want to understand this, you really need to try harder to read what is written.

I don’t know exactly how performance based SR adjustment works, since Blizzard hasn’t said and my attempts to reverse engineer it have not been successful. One possible answer, though, is that both SR and MMR get performance modifiers independently, but if SR gets too far away from MMR, SR gets pushed in the direction of MMR (this second part has been verified, for decay and the match leaving penalty).

I give my arguments on which statements on SR vs MMR are more technically correct in the original post, so I will not repeat them here.

If Jeff’s statements are to be ignored, then the conversation is over. Without Jeff’s statements, proving the actual existence of MMR would be difficult (and practically impossible after decay is removed from the game next patch).

The streaks and stomps are are fully consistent with a non-rigged system, with only one match-making-rating, and I have done substantial analysis.

Overwatch Forums and https://www.reddit.com/r/OverwatchUniversity/comments/aatezy/why_match_quality_is_frequently_poor/

Note, however, that role queue should reduce streak length, SR swing, and stomp frequency.

Mandalorian-11203 · August 10, 2019, 6:52am

And if a players MMR is “wrong” and too high… then what? And who or what decides that a players MMR is “wrong?”

But how does the MMR get “too high” (or low). I mean, that player PLAYED and got a certain MMR.

Also, from what i’ve observed over 3 years. You don’t get put with stronger opponents AND allies. You just get worse allies. I’ll climb with decent teammates for a week, and then suddenly the good players disappear. No one can make a comp. People throw at the start of a match. And the overall quality of teammates drops SO drastically and THAT is the reason my win rate approaches 50%.

RexOverwatch-1555 · August 10, 2019, 7:00am

you can’t fake a win, but you can always throw a match

all 11 people in your match have no permanent influence over your SR.

this is probably observational biased to be frank

ladder isn’t about making a comp, it’s about navigating the chaos

the reason your win rate approaches 50% is because that is how ranking/ladder systems work. The more you win, the harder the matches get. The more you lose the easier the matches get.

Mandalorian-11203 · August 10, 2019, 8:08am

Hmm, trying to figure out why the games at 2400 on the way UP to 2600 were so easy, yet now i’m seeing impossible games at 2300 when the system seemed to decided for me that i didn’t belong at 2600. It certainly wasn’t the enemy team at 2600. But the good comps, good coms, and decent people that i was getting as teammates on the way up sure disappeared. Is everyone at a higher rank all the sudden a bunch of toxic players who start throwing at the start of a match for 10 games in a row?

All i can say is, what you describe is NOT what i’m seeing.

Is the game intended to be HARDER at 2100 than 2600? That’s what i see. Oh, until suddenly i’m playing with a bunch of toxic fools and get to start the cycle all over again.

What i’m seeing is, you rank up and have fun games until suddenly you deserve a bunch of instantly toxic teammates, seemingly because the system itself has decided you don’t belong up there. I’m not seeing tougher enemies, or better teammates at a higher rank. I’m seeing just… toxic teammates all day, every day until my win rate is 50%.

And since that is something that’s hard to prove without having you actually play 100’s of games with me. I guess my story is all i’ve got. Same with others who’ve said something similar. I believe them. I don’t think gamers are idiots who can’t tell a good teammate from a bad teammate. The same way a basketball team made up of toxic players who can’t even shoot straight would be easy to spot after you’ve just played with a bunch of good, decent players for a few days.