“Matchmaking is rigged.” I’ve heard this from players at every rank in every competitive game. It’s one of the most common complaints in gaming, and it’s almost always wrong — but the frustration behind it is real and worth understanding.

Modern matchmaking systems are sophisticated pieces of engineering that use AI and machine learning to create fair, enjoyable games. They work better than most players think. They also have genuine limitations that explain why some matches feel terrible.

Here’s how they actually work.

The basics: Elo and its descendants

Most matchmaking systems are built on some version of the Elo rating system, originally designed for chess. The concept is simple: every player has a numerical rating. When you beat someone rated higher than you, your rating goes up more. When you lose to someone rated lower, it goes down more.

Modern systems have evolved well beyond basic Elo. VALORANT uses a system called RR (Ranked Rating) backed by a hidden MMR (Matchmaking Rating). League of Legends uses LP (League Points) visible to players, with a hidden MMR underneath. Overwatch 2, CS2, and other titles have their own variations.

The hidden MMR is where the AI comes in. Instead of a single number, modern systems track multiple dimensions of player performance. Your MMR might account for:

• Win/loss record (the primary factor)
• Individual performance metrics (damage, utility usage, objective play)
• Consistency (a player who varies wildly between games is rated differently from a consistent one)
• Role-specific performance (some systems rate you differently depending on what role you play)

The AI component learns from millions of matches which combinations of these factors best predict match outcomes, and it continuously adjusts how each factor is weighted.

How matches are created

When you queue for a competitive match, the system is trying to create a game where both teams have roughly a 50 percent chance of winning. It does this by assembling teams with similar aggregate MMR, while also considering:

Queue time. The longer you wait, the wider the system searches. After a short queue, your match will be closely balanced. After a long queue, the system relaxes its standards to get you into a game. This is why late-night matches in smaller regions (like OCE) tend to be less balanced — the player pool is smaller, so the system has to compromise.

Party size. Groups of players queuing together complicate matchmaking. A five-stack of friends at different skill levels is hard to match against anything. Most systems handle this by averaging the group’s MMR, which means someone in the party will be playing above their level and someone else below.

Regional considerations. In Australia, the matchmaking pool is small enough that the system regularly makes compromises that wouldn’t be necessary on larger servers. Wider rank ranges, longer queue times, or routing to SEA servers are all tradeoffs the system makes for Oceanic players.

Why it feels rigged (but isn’t)

The most common complaint is “forced 50 percent win rate” — the belief that the system deliberately gives you bad teammates after a winning streak to drag your win rate back to 50 percent.

This isn’t how it works, but it feels true because of how MMR adjustment operates. When you win several games in a row, your hidden MMR increases. The system then places you in harder matches to test whether your new rating is accurate. Harder matches mean you’re more likely to lose. This feels like the system punishing you for winning, but it’s actually the system calibrating your skill level.

The alternative — keeping you in the same skill bracket after a winning streak — would mean you’d stomp easy games indefinitely. That’s not fun for anyone.

Another source of frustration is teammate variance. In a team game, your individual performance is only partially responsible for the outcome. Even in a perfectly balanced match, you might get teammates who have a bad game, don’t communicate, or play a style that clashes with yours. Over a large sample of games, this averages out. In any individual game, it can feel terrible.

What AI is improving

Recent advancements in matchmaking AI focus on factors beyond raw skill:

Behavioural matching. Some systems now factor in player behaviour — communication style, toxicity reports, and play style preferences. The goal is to match players who will have a good experience together, not just players of similar skill. This is still experimental, but early results from studios that have tried it show improved player satisfaction.

Role-aware matching. Instead of treating every player on a team as interchangeable, newer systems consider what roles players prefer and how those roles interact. A team of five players who all want to play the same role is worse than a team with complementary preferences, even if the aggregate MMR is identical.

Predictive modelling. AI that predicts not just who will win, but how the match will feel. A 50/50 match where one team stomps the first half and the other team comes back is a great experience. A 50/50 match where both teams play passively and the outcome is decided by one lucky play is boring. Systems are starting to optimise for match quality, not just balance.

For the Australian context, matchmaking AI could help with our specific challenge: small player pools. Smarter systems that create better matches from limited options would disproportionately benefit regions like OCE where the current systems already stretch to fill lobbies.

The technology companies helping build these systems — including AI consultants Brisbane working with gaming clients — are focused on making matchmaking feel fair even when the numbers are tight.

The bottom line

Matchmaking isn’t perfect. It can’t be, given the constraints it operates under. But it’s significantly better than the alternative — manual server browsers where skilled players dominate every lobby and new players get destroyed.

Next time you get a bad match, remember: the system is trying. It’s working with limited information, a finite player pool, and the impossible task of making every game feel fair. Most of the time, it succeeds. The times it doesn’t are just more memorable.