# What K factor should be used if two players are in different K factor brackets?

Apparently, when calculating a player's USCF rating difference after a win or loss, the K factor that is used is based on the player's rating. Now, I was always under the impression that two players with ratings `a` and `b` would play each other and that a difference in score is awarded to both of them such that their ratings after the match would be `a+d` and `b-d` (assuming player A won the match).

But let's say you have a player at 2075 (who would have a K factor of 32) and another player at 2125 (who would have a K factor of 24). Which K factor would be used? I could see a case for either of 32, 24, or even 28 being used in this case. What does USCF do here?

• I don't know USCF rules exactly, but if this turns out to be anything other than that each player is adjusted according to his own K-factor (so they get different changes) then I'd be amazed. – RemcoGerlich Dec 15 '13 at 19:50
• Wouldn't that mean that one player earns a different amount than the other player loses? I'd expect that to be an invariant. – corsiKa Dec 16 '13 at 1:19
• RemcoGerlich is correct, and yes, it does mean that. – dfan Dec 16 '13 at 21:21
• The USCF system described in the linked Wikipedia article is out of date. The K factor is no longer simply 16, 24, or 32 selected by a simple table lookup based on rating. It is now calculated based on the effective number of games played and number of games in the current event. Effective number of games is based on actual number of games and current rating. Full details here: glicko.net/ratings/rating.system.pdf – tzs Dec 17 '13 at 3:43