The problem of vastly unequal ratings at different time controls

I’m currently playing one correspondence game in which my opponent got a four stone handicap even though his overall strength is actually two stones above mine, and another game where my opponent got six stones even though he’s the same strength as I am overall. The reason these one-sided games came about is that their correspondence ratings are WAY lower than their ratings in the other time controls where they’ve played lots of games. I’m not sure how this could happen, because I see that I have nominal ratings for the time controls I’ve never played, and they’re fairly close (about 100 points below) my correspondence rating.

Could it be that they played one or two correspondence games way back when their ratings were much lower, so their correspondence ratings got stuck there while they rose in rank overall?

In any case, is there anything you can do about this? Maybe default to a calculated rating whenever a player has played fewer than N games in the last M months at a given time control? This can’t be fun for my two opponents, either: they have to play tons of meaningless, utterly unchallenging games at the slowest of all the time controls just to get their correspondence ratings up to their actual strength.



I agree. This is more fundamental then one can think. Let’s examine currently running “Monthly Simultaneous McMahon 2016-03-16 17:00”. I have a friend (16k) and he got into group were all players are 9-10k because their correspondence game ranks are way below overall rank.

He is downdog now and doesn’t have any motivation.

I think applying a weighting curve instead of hard cutoffs, the way it’s done with the smaller board sizes, could work better. Something like the diagram below (excuse my bad drawing)

Correspondence games would have to have either a fixed weight from (and to) other games or something like “thinking time” recorded and used instead of the actual time controls.

Players coming back from a hiatus would have their present rank weighed less, maybe exponentially decaying over ~3 months after which they become provisional again.

IMO, only the “overall” rank needs to be shown. It’ll be less messy, and time dependant and board size dependant performance differences could be accounted for silently during rating. A 3D graph of rank vs time control and board size would be pretty cool though.

I am unfortunately out of uservoice votes, but maybe someone could suggest it at

1 Like

I don’t understand your post.

Do you mean that correspondence should be counted for blitz rank, albeit with a very small weight?

Yes, that’s right. I think that because playing strength between time settings are correlated, albeit more weakly than the ones inside the same time settings, so they’d still make for a useful guide if the user doesn’t play at one of the time settings often.

1 Like

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.