When a game says it is finding you a "fair match", it is usually doing one simple thing: guessing how good you are as a single number, then looking for opponents with a similar number. That number is called a rating, an MMR (matchmaking rating) or an Elo score, and the idea behind most of them is surprisingly small.
Elo was invented for chess by the physicist Arpad Elo, and the world chess federation adopted it in 1970. Many games use it directly or a close cousin of it. Others use newer systems such as Glicko or Microsoft's TrueSkill, which add extra bookkeeping, but they all start from the same question: how likely is this player to beat that one?
In Elo, only the gap between two ratings matters, never the ratings themselves. Equal ratings mean a 50% chance each. A gap of 100 points gives the stronger player about a 64% chance. A gap of 400 points gives them about 91%. The maths is one line: the expected score is 1 / (1 + 10^(gap / 400)), with the gap measured against the stronger player.
Notice what that means for you. If a game matches you against someone 400 points above you, you are meant to lose about nine times out of ten. That is not the game being unfair, it is the game working as designed.
After each match your rating changes by K × (result − expected). The result is 1 for a win and 0 for a loss, and K is a "how fast should I react" setting. If you were expected to win 90% of the time and you win, you barely move. If you lose, you drop a lot. Beating someone much better than you is the big payday.
The play area below lets you feel this. Set a gap, then press Win or Lose and watch the change.
A new account has no history, so the game must guess. Hit Reset in the demo and play a few batches: the line climbs quickly at first, then flattens near your true level. That is why early matches can feel very one-sided. Many games also use a higher K, or an uncertainty value as Glicko and TrueSkill do, for new players so they settle faster.
The reverse also holds. Once you are near your true level you win roughly half your matches, and your rating just jitters around. Winning 50% is not a sign you are stuck, it is the matchmaker doing its job.
A rating only knows results. It does not know about your ping, your team's mix of roles or whether you queued at 4 a.m. Real games add those on top: they widen the search the longer you wait, try to keep ping low, and balance teams so the average ratings match. Each trade-off is a choice between a fairer match and a longer wait.
Team games add one more wrinkle: your rating is estimated from the team's result, so a single match tells the system less about you than a one-on-one would. That is why climbing in a team game can feel slower and noisier.
A useful habit is to compare your results with what the rating predicts. If you are rated about the same as your usual opponents and win around half your games, the system has you placed correctly. If you win far more than half over a long stretch, your rating is too low and is about to rise. If you lose far more, it is too high and will fall. Either way the number is catching up with you, not punishing you.
Treat your rating as a rolling estimate, not a verdict. Judge progress over dozens of matches, not three. A short losing streak is normal even at a fair rating, and a steady climb over weeks is the real signal that you are improving. Specific systems differ in detail and most publishers do not publish their exact formulas, so use the demo as the idea, not as the exact rules of any one game.