What SBMM is and why so many players hate it

What SBMM actually does

SBMM stands for skill-based matchmaking. It estimates how well you play, then looks for teammates and opponents within a reasonable range. The game can use it in public playlists without showing Bronze, Gold or Diamond ranks. It does not mean an algorithm has decided that you must lose the next match.

The trade-off appears quickly. Narrower skill gaps also mean fewer relaxed matches where you can test any weapon and win without paying much attention. That constant pressure protects newcomers, but it can make a casual mode feel like ranked play without the rank or rewards.

The game never knows your ability with certainty: it estimates your level from previous results and whatever signals the developer has chosen. Activision says Call of Duty updates that value after every match according to relative performance. Microsoft's TrueSkill 2 combines results with information such as experience, party composition and individual performance to calculate what should happen in future Halo and Gears matches. An estimate can be useful over hundreds of games and still be wrong about the next one.

Skill is only one part of the search

If lobby, ping and MMR are starting to blur together, jump to the online play, ranked and matchmaking section of our gaming dictionary, which defines each term before you go any further.

Using SBMM does not require a game to sacrifice connection quality. Activision's Call of Duty paper puts connection first and sums up that priority as 'Ping is King'. Search time, platform, input device, voice chat, recently played maps and the available playlist pool can all affect the result too.

Skill has to fit around those constraints, and the game gradually expands the acceptable range when it cannot find enough suitable people. Apex Legends describes the same trade-off between skill, waiting time and latency. A search at 4 pm in a busy region therefore draws from a very different pool than a niche playlist at 3 am.

Ping is the time data needs to travel to the server and back, adding to delay from the controller, game and display. Our guide to input lag and how to reduce it helps distinguish a network problem from slow response elsewhere in the setup.

Matchmaking and team balancing are not identical either, because the game may assemble a reasonable lobby and only then split it into two sides. One disconnect, a poor mix of roles, a coordinated party or ordinary chance can still turn that match into a blowout.

How it stops newcomers from leaving

Picture your first afternoon with a shooter. You are still looking for the weapon-swap button while the opponent knows every corner after a thousand hours. If that gap repeats match after match, you do not learn much: you spawn, die and eventually close the game. SBMM's first job is to prevent those extremes while somebody learns the rules.

Activision tested that effect by reducing the weight of skill in several Call of Duty experiments. The company says most groups played for less time and quit more matches, with the clearest impact at low and middle skill levels. It also argues that meaningful skill matching improves the experience for 80% to 90% of the population it studied. Those are aggregate results, not a promise that every individual match will be close or enjoyable.

Keeping newcomers helps everybody else too. If they leave, the pool shrinks, average skill rises and yesterday's middle group becomes the weakest tier in most lobbies. More players support reasonable opponents, quieter playlists and nearby servers. The cost is just as real: experienced players get fewer matches in which they can switch off and more games demanding attention from the opening minute.

Why every match can feel like a final

The goal is not to force your account to remain at a 50% win rate. Matchmaking tries to create a game where its estimates do not give either team an overwhelming advantage before play begins. If those estimates remain accurate over many matches, a player's win rate will tend to move away from the extremes. The system does not need to manufacture one specific loss to produce that pattern.

TrueSkill 2 shows why rating accuracy matters. Microsoft reports that it predicted the winning team in 68% of the Halo 5 matches it analysed, compared with 52% for the previous model. A better rating lets matchmaking avoid players whose estimated levels are too far apart and split a lobby into teams with more similar predicted chances. That is the causal step between a more accurate model and closer groups.

Mixed-skill parties create another difficult choice. Protecting the newcomer may let their experienced friends dominate the lobby, while matching around the strongest member can leave the newcomer barely able to take part. Neither result satisfies the whole group.

Reverse boosting adds a further problem. A player can deliberately perform badly to lower the estimate and reach easier lobbies later. The system has to recognise that manipulation without punishing somebody who simply had a rough evening. Limited information encourages suspicion, especially around competitive launches such as the Modern Warfare 4 beta, when every lobby change can feel personal.

SBMM is not the same as EOMM

EOMM stands for engagement-optimized matchmaking. Instead of focusing only on skill balance, the model aims to increase the chance that somebody will keep playing in future. Public research has explored that idea, including a 2017 academic paper.

The paper does not prove that any particular game uses EOMM, much less that a game secretly alternates wins and losses for each player. Those are separate claims and require their own evidence.

Studios should still explain what their systems prioritise, how they treat mixed parties, when they expand a search and whether casual playlists follow the same rules as ranked modes. Without those answers, every unexpected loss can appear to confirm whatever theory a player already holds.

Before blaming SBMM, separate three possible problems. Poor connection points to ping. A one-sided match between similarly skilled players may come from the way the lobby was divided. Consistently meeting opponents far above or below your level raises a question about the skill estimate itself. Frustration grows when the game blends those experiences together and never explains which decision it made.

SBMM solves a real problem, but it needs limits

SBMM is not a conspiracy for deciding who loses. It is an imperfect answer to one concrete problem: stopping new players from becoming target practice for people with hundreds of hours behind them. Removing it entirely may make some veterans' evenings easier while making everybody else's first week much worse.

Solving a real problem does not make every implementation defensible. A good system should preserve a stable connection, allow wider skill gaps in casual modes than in ranked play and explain how it handles mixed parties. If every public match demands the effort of a final, the developer has protected the entrance to the game by removing the space to relax inside it.

Our verdict is straightforward: skill matching is worthwhile when it reduces extreme blowouts without ruining ping or turning the whole game into ranked play. The strongest criticism is not that a game manufactures losses, but that too many studios hide where they set that balance. The less they explain, the easier it becomes for any bad evening to feel manipulated.