[Apex Legends developers, like all competitive games out there, have to deal with players' perceptions of what is at stake, along with the reality they see behind the scenes. a persistent source of frustration for the Apex community is the game's matchmaking system, specifically about how to measure skill and form squads and matches around it.

The list of various issues raised by players would be long, but a typical example is that skilled players sometimes feel that they are intentionally matched with less skilled players in an attempt to create balanced matches, and vice versa.

This problem has been going on basically ever since the game was released, but the developer, Respawn, has now posted what can only be described as a comprehensive response to the community. It is quite voluminous and goes into great detail about how the game's matchmaking works, the changes that have been made recently, the problems that the developer acknowledges, and what is being done to fix them.

If you are an Apex head, it is probably best to read the whole thing (open in a new tab) (have a cup of tea first):

According to Respawn, there are three factors to consider in matchmaking: a progression system visible to players, hidden skill ratings and the matching system. The progression system may seem a bit strange, but according to Respawn, it affects not only player behavior, but also player engagement and can "strongly influence players' perception of how fair a match is. The studio also states that account level does not correlate with skill, which is also a factor: "Sometimes low-account players are highly skilled (as most of us here are...). Sometimes high-level players are the potatoes."

Skill ratings are simple numbers, invisible to the player. A chart is then displayed showing one view of the skill level distribution. The main takeaway from this is that the vast majority of players are in the middle, with few at the extremes. There is also what Respawn calls a "bucket of pieces" approach, where players are divided into five groupings (0-2, 2-4, etc.) based on these ratings, and a "fairly wide" range for unranked matchmaking.

The matching system uses skill ratings to group players, obviously based on who is looking for a match at that particular time. Once a group is chosen, another problem arises.

Respawn explains in detail how the algorithm attempts to form fair teams, and this is where it comes in. In non-ranked matches, the game will form teams with the closest average skill, but it does things differently in ranked matches.

"In a fully competitive environment such as a ranked match, the emphasis is more on having teammates with the same skills and less "carry" than on a completely fair match. Note: We are effectively stacking teams here, but with a limited number of players to pick from. Because of the inherent randomness and chaos of battle royale, we believe this is the right trade-off."

There is more on how the game calculates team ratings, but the key result is the one above; Respawn further states that it is "inevitable" that there will be a mix of pre-made squad sizes and squad sizes that the game forms from random.

As a solution, Respawn has a new algorithm that will be implemented in early 2023. The new system "does a better job of complementing the competitive advantage that pre-made duos and trios have. We have been experimenting with a new system that is scientifically validated to quantify this advantage, continuously learning and updating over time". The core of this new method is to recognize the slight advantage of being a premade duo or trio and add this to the skill measurement of a given team.

This new system will also replace the various ways that Apex Legends is currently matchmaking across modes. Once this new system is in place, "all matchmaking will use the same skill evaluation technique, but each mode will use different tuning values based on what works best for that mode."

And remember the skill buckets: there will only be more buckets.

"The new system is much finer and has more buckets. This allows the matching algorithm to make better decisions when placing players in matches."

Respawn provides some mathematical examples of how it thinks this new system will measure skill in a more "meaningful" way and be smarter about how it uses all the buckets in different time frames (it also predicts other buckets coming and incorporates this into its calculation of incorporate: quantum buckets). According to the studio, live tests have shown that this system makes "matches tighter, and latency remains almost the same," and the graphs prove it.

The old system will be phased out across the region to allow Respawn to monitor the launch of the new system and address any unforeseen issues, and once the replacement is complete, the matchmaking team will further elaborate on "additional modifications" to further account for the solo/duo/trio nature of premades They will begin to consider.

Big news for Apex Legends enthusiasts, but in addition to the details of the new system, Respawn has also answered some more general community questions. On the topic of why skilled players end up in matches, he said this is due to the dynamic adjustment of skill ratings. This works both ways, so if you are on a losing streak, the match should gradually become "easier". If the opponent is a higher ranked player, it is possible that he "just happens to be out of shape and is on a losing streak". Other factors include the distribution of skills during the match (which will apparently be "reduced" in future changes) and the fact that more skilled players can bring weaker players into the match.

Does the game give good players bad teammates?

"No, it does not intentionally give good players bad teammates," says Respawn, before explaining that "in a way it does. 'If you are one of the stronger players in a particular match, you are certainly more likely to be paired with a player weaker than you.' This is because the tail end of the distribution is less populated, and matchmaking in this zone requires incredibly long matchmaking times, incredibly large skill differences, or a mix between the two. He also noted that "skilled players tend to squad up, which exacerbates the problem."

Respawn acknowledges that this may effectively feel like "big brother matchmaking," and the new system aims to mitigate the impact through "tighter matching that reduces these problems."

Finally, they also address the question of whether the matchmaking system is built with the goal of optimizing retention and engagement: that is, it is built to keep players playing Apex Legends as long as possible. The studio notes that a major difficulty is trying to create "the most fun matches" when "fun cannot really be measured."

Respawn uses retention as a criterion, reasoning that "players are more likely to stick around longer if they have fun." However, they have made it clear that "we will never build an algorithm that directly optimizes retention (absolutely no engagement - convincing people to play an extra hour a day when they would normally be doing something else is not good for us or for you)."

The continued success of Apex Legends has undoubtedly transformed Respawn as a studio. Whereas many others have tried and failed to do live services right, Respawn has succeeded in attracting and retaining a huge player base by offering a constant stream of new content and ways to play. It has also changed the way it engages with the community, and "Apex" is arguably one of the best-in-class examples of a major studio being as responsive and transparent as possible. This long and comprehensive explanation of the game's matchmaking principles, the problems inherent in them, and how the studio is working to improve is exemplary communication. Even if every time you lose a match, you will feel that it is the game's fault, because "the game is not a game, it is a game.

.