How chess ratings really work

Every online player has a rating, obsesses over it a little, and rarely knows what it actually means. Here's the plain-English version — enough to understand your number without a statistics degree.

A rating is a prediction, not a grade

Your rating is a single number that predicts your results against other rated players. If you're rated 1500 and your opponent is 1500, the system expects you each to score about half. Beat a higher-rated player and you gain more; lose to a lower-rated one and you drop more. That's the whole idea: the number moves until it predicts your results well.

Elo and Glicko

The classic system is Elo — simple, and still used everywhere. Most online sites use Glicko (or Glicko-2), which adds a 'rating deviation': a measure of how sure the system is about your number. New accounts and players who haven't played in a while have a high deviation, so their rating moves in big jumps until it settles. That's why your first twenty games feel so volatile.

Why it bounces around

Ratings are noisy because chess results are noisy — you'll beat stronger players on good days and lose to weaker ones on bad days. A single session can swing your number 50 points either way and mean almost nothing. Look at the trend over dozens of games, not the result of the last one.

Different pools don't compare

A 1600 on one site is not a 1600 on another, and neither matches an over-the-board national rating. Each pool is its own scale. Compare yourself to your own past, not to a friend's number on a different site.

Chase improvement, not points

The players who climb fastest aren't watching their rating — they're reviewing losses, drilling tactics, and learning endgames. Do that and the number follows. Stare at the number and you'll just play scared. Treat your rating as a thermometer, not a goal.