基于技能的對局配對
基于技能 combat 匹配（Skill-based matchmaking）将测定能力相似的玩家分组，以保持比赛的竞争性。它在排队时间与比赛质量之间进行权衡：更窄的技能区间能带来更好的比赛，但等待时间也更长。如何设定这个区间，以及在玩家等待时如何快速放宽限制，是整个设计的核心问题。
又称
基于技能的比拼匹配 (SBMM)
，
What Does Skill-Based Matchmaking Do Inside a Matchmaker?
A matchmaker turns a queue of players into matches. Each player's request carries attributes: ping to each server location, party size, input device, game mode, and a skill rating. Rules decide which requests can share a match.
Skill-based matchmaking is the rule that compares skill. It's one rule among several, and its priority depends on many variables, from the player count to the type of game.
Call of Duty is the best documented case. Activision lists ping, time to match, playlist variety, input device and voice chat alongside skill, and calls connection "the most critical and heavily weighted factor" in the process (Activision, April 2024). In its design, when rules pull against each other, skill targets loosen faster than connection targets as a search runs longer. Our breakdown of Call of Duty's skill-based matchmaking walks through each factor.
The rating behind the rule is its own subject. How it's calculated and updated is covered under MMR and ELO Matchmaking.
Why Does Skill-Based Matchmaking Matter?
Much of what developers hear about SBMM comes from players asking to remove it. Activision tested what happens when it's loosened.
In early 2024, it relaxed skill constraints for half of Modern Warfare III's players in North America. The returning-player rate fell significantly for 90% of players, and the quit rate rose for 80%. Only the top 10% came back in greater numbers (Activision, 2024). With skill in matchmaking, Activision reports that 80 to 90% of players get a better end-of-match placement.
The design philosophy behind SBMM is to benefit the majority of a player base: fewer one-sided matches for the players most likely to lose them, and so a pool that stays large enough to match everyone well.
Not everyone agrees with that trade. Strong players lose the easy matches they used to get, and casual playlists can feel as tense as ranked ones. How a studio weighs those views is a design choice, and the Call of Duty data is one input to it.
How Tight Should the Skill Band Be?
It depends on how many players are searching. A matchmaker typically looks for players within a narrow skill range first, then widens the range as the wait grows. Two settings define it: how narrow the band starts, and how fast it widens.
Game design and data point toward matching the band to the pool. A large shooter with thousands of players searching in each region can often hold a tight band and still fill matches in seconds. A smaller game with the same band tends to wait, then widen to a similar match anyway, only later. Every extra split by region, mode or platform shrinks each queue further, the cost regional matchmaking runs into.
The band can reach past the first match, too. A tight band can slow backfill, because a replacement player has to fit the same range, and servers can run with empty slots while they wait.
When Does Skill-Based Matchmaking Matter Less?
Skill tends to matter most where players compete directly and the pool is large. Design practice and published data point to a lighter touch, or a looser band, in a few common cases:
Small or new player pools. With few players searching, a tight band can empty the queue, so connection quality often leads, with skill as a loose preference.
Launch. New players have little match history, so their ratings are still settling.
Co-op and PvE. With no opponents to balance, party makeup and connection often matter more.
Social and party play. Friends of very different skill play together, and any single rating for the party is a compromise.
Ranked modes sit at the other end. Call of Duty's Ranked Play pairs a visible Skill Rating with a hidden skill measure, and its median search time is twice as long as casual modes (Call of Duty's ranked matchmaking). Players accept the wait because the result counts.
Edgegap's matchmaker expresses these choices as configuration: latency rules, a rank or skill rule beside them, and expansions that loosen each rule on its own timer. The competitive game examples in our documentation show both a rank-based and a skill-difference profile.
来自我们的赞助商（其实就是我们自己！）的寄语
当其中一名玩家就住在数据中心旁边时，比赛就是不公平的。Edgegap 的边缘云可按需利用多达 615 个以上的分布式节点，根据每场比赛中所有玩家的连接位置来部署服务器，而不是固定在某个区域。在 Edgegap 的 1v1 研究中，这让对手之间的公平性差距缩小了 28%。
Edgegap 的观点（仅代表个人意见，姑且听之即可！）
Find the Balance, Then Keep Tuning It
Skill-based matchmaking is part of matchmaking, not the whole of it. How much weight it deserves depends on the game: how many players are searching, the genre, the mode, and what players came for. A ranked shooter and a co-op raid won't land in the same place, and neither will the same game at launch and a year later.
What the published data offers is an order of trade-offs. Activision weighs connection most heavily and lets skill loosen sooner as a search runs long. An uneven match still plays. A match on a poor connection often doesn't.
Whatever balance a game starts with, the live data will move it. That's why Edgegap's matchmaker keeps latency, skill and rank as separate rules, each with its own expansion timer: the balance is configuration a team can tune from what its players actually do, not code to rewrite.
，
