Being placed into a match is the output of a system solving three problems at once, and the three pull against each other. Every match is a compromise among them.
Skill is an estimate with uncertainty attached
Systems maintain a rating for each player, along with a measure of how confident they are in it. New accounts carry high uncertainty and move rapidly.
Results update the estimate, with an unexpected outcome moving it further than an expected one. That is why beating a stronger opponent is worth more.
Confidence grows with games played, which makes ratings stabilize over time and explains why established players climb more slowly than newcomers.
Connection quality is a hard constraint
A match between well-matched players on distant servers plays badly regardless of how even the skill pairing is, so latency limits are enforced before skill is considered.
This restricts the eligible pool to players within a geographic range, and the restriction tightens in regions with fewer players online.
Players in smaller regions therefore face a genuine tradeoff between waiting, playing someone poorly matched, or accepting a worse connection.
Wait time is the pressure that loosens everything
Search criteria widen as a player waits, accepting larger skill gaps and higher latency the longer the queue runs.
The system is tuned around what players will tolerate, and most will accept an imperfect match over an extended wait.
This is why matches at unusual hours or high ratings feel less balanced. The pool is thin and the criteria have widened to fill it.
Teams are balanced by average, which hides variance
Team games generally balance the sum or average of ratings across sides rather than pairing individuals, since exact pairing would make matches nearly impossible to form.
A team containing one very strong and several weak players can be nominally equal to an evenly matched opponent while playing nothing like it.
Groups queuing together compound this, because coordination is worth more than the ratings capture, which is why many systems penalize or separate large parties.
Perception diverges from the design target
A well-tuned system produces close matches most of the time, which by design means winning roughly half of them.
Players experience that as frustrating even when the system is working exactly as intended, since a run of close losses feels like unfairness rather than balance.
Developers respond by exposing more information about how placement works, on the reasoning that a visible mechanism attracts less suspicion than an invisible one.