0

It seems that when one loses a game against a compatibly rated opponent, the (USCF) rating point decrease is more significant than when one scores a win.

For example, between two 'C' players, losing an over-the-board game may cost a player 12 points, whereas wining, the increase in rating is 7 points.

This is an odd phenomenon in the Elo system. Is there an explanation?

2
  • 8
    This can't be true. Your example implies that everyone's rating will quickly deflate. I think winning a losing is symmetrical. Commented Sep 5, 2015 at 21:22
  • 3
    Absent other factors like provisional ratings, the USCF system does not work this way. If you can point to a rating report for a tournament that you think demonstrates this asymmetry, that would help clarify what is actually going on.
    – dfan
    Commented Sep 9, 2015 at 1:50

4 Answers 4

6

I suspect it's a result of you not really knowing the rating of the player you're playing. Posted rating for a player in a tournament is typically already out of date. You can see it in more detail by checking the actual rating report of the event you played in (http://www.uschess.org/datapage/event-search.php) and you can see what the real rating of the player you played was at the start of the event. Keep in mind it changes with every round.

2
  • Rating's aren't adjusted round by round. They ARE, however, affected by how your opponent did in the tournament. If your opponent's rating went down in the tournament, then your rating is calculated based on you playing that lower rated opponent.
    – D M
    Commented Feb 10, 2018 at 17:37
  • Weeeelllllll......Ratings are not recalculated every round, but the effective rating of the player you're playing changes every round, as the previous round results play into the calculation made after the tournament is submitted. Remember, there's absolutely no guarantee the player you're playing will play another round in the event.
    – Arlen
    Commented Feb 18, 2022 at 18:27
3

How much rating you lose or gain depend (besides your and your opponents rating) on the K-factor. If your opponent has a different K-factor than you do, the gain/loss of rating can be asymmetrical.

In the Fide Elo system the K-factor depends on rating and age. Apparently the USCF has a K-factor that is dependent on the number of games played overall and in the current tournament.

So, that is probably the reason for the discrepancies you noticed. But it has nothing to do with asymmetry between winning and losing.

1
  • 1
    This is the best answer.
    – AlanW
    Commented Dec 1, 2017 at 19:16
-2

ELO ratings is all about statistics. When you win against a weaker player you gain less points then you lose when you lose against that player. You can also lose points when you draw against a weaker player.

-2

Due to the huge upsurge in rated players, and the lesser skill level required to become one, we see a player becoming rated every other day. If USCF did not implement this imbalance, Fabiano Caruana might get rated at 3000+ as rating is being 'added' into the system for every new rated. This imbalance 'subtracts' rating from the system to keep it more or less even.


So if a new player gets rated at say 1800, due to less of rating the 1800 he has added to USCF will eventually drain out.

2
  • 1
    Simply having more players doesn't create inflation. And there is no such imbalance; if two players with the same rating and same K play each other, the loss will equal the gain.
    – D M
    Commented Aug 23, 2018 at 21:51
  • all it takes to be a rated player is the entry fee to a tournament. it has nothing to do with how many players there are. new players tend to pull down ratings not inflate them as most players improve fast but decline slowly assuming they are still playing. the k factor helps some.
    – yobamamama
    Commented Jan 9, 2020 at 19:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.