How FIDE Ratings Actually Get Recalculated
Magnus Carlsen's 18-point drop after Norway Chess and the UzChess Cup, see our context piece on why it matters less than the headline suggests, raised an obvious question: how does one tournament move a top player's rating by double digits? The mechanism is simpler than it looks, and understanding it changes how you read every rating-list headline afterward.
The core idea: expected score versus actual score
Every FIDE-rated game has an expected result baked in before a move is played, calculated from the rating gap between the two players. Beat someone rated 200 points below you, and the system expected that anyway, your rating barely moves. Lose to them, and the system is hugely surprised, and your rating moves a lot. Carlsen's rating drop was not one catastrophic result, it was several results across two events that each scored slightly below what his rating expected, compounding into 18 points.
K-factor: the volume knob
The size of each individual rating swing is scaled by a K-factor, effectively a volume knob on how reactive a player's rating is to a single result. Lower-rated and newer players get a higher K-factor, so their rating reflects recent form quickly. Elite, established players like Carlsen have a lower K-factor, their ratings move less per game, which is precisely why an 18-point move across two entire events, rather than one bad game, stands out as unusual for him specifically.
Why a draw can still cost you points
This is the detail that confuses most club players: a draw against a much lower-rated opponent can still cost rating points, because a draw scored 0.5 when the system expected close to 1.0 counts as underperformance. That is exactly the kind of result, not a single loss, that can quietly account for several points of an elite player's drop across a long event.
Try it yourself
Next time a rating list headline says a player "dropped" or "gained" a specific number, ask two questions: how many games produced that number, and against what strength of opposition? An 18-point drop from one bad game against a peer is a different story than the same 18 points spread across nine games in two strong events. The number alone never tells you which.