I've noticed a slight calculation error when tracking my own rating. The difference between my rating calculation and RHP's is only one point (at this stage).
Is a person's rating truncated to be a whole number? That, which is displayed. Or, are there hidden decimals that we don't ever see, and that get included in one's rating, and calculated against an opponent?
Kind of like that 1/2 penny in Jumping Jack Flash, with Whoopie.
Originally posted by nilocWho cares?
I've noticed a slight calculation error when tracking my own rating. The difference between my rating calculation and RHP's is only one point (at this stage).
Is a person's rating truncated to be a whole number? That, which is displayed. Or, are there hidden decimals that we don't ever see, and that get included in one's rating, and calculated against an opponent?
Kind of like that 1/2 penny in Jumping Jack Flash, with Whoopie.
Originally posted by nilocWell, I'm not gonna lose any sleep over any calculation truncations. I'm no math geek, but my gut tells me that it wouldn't make much difference. Lets say your rating is temporarily below your real skill level. Your rating will still tend to converge upward towards your real skill level. If there are any calculation truncations, then the convergence might be slightly slower, but my guess is that the difference in convergence rates would be hardly noticable.
Apparently, I do! 😏
If there is a small error with a limited number of sample points, is there a larger error as the sample pool gets larger? Or, is it just limited to a +/- 1 point variance?
As far as the sample pool issue is concerned, elo ratings are relative, and the distribution curve is assumed to be centered on an arbitrary elo rating (about 1500 for the USCF). I don't think rating calculation truncations would have any effect on this distribution curve.
Originally posted by epic0002Cardinal Dude
dude its online chess...you can have one of my points if you shut up about it
It's only a question, Chill! 😀
Besides, roundoff errors can have huge impacts in real life. I'm just curious if the programmers are aware. Since memory became cheap, nobody cares about the details of programer gluttony, that is until you find your 401K has been squandered away by a "computer glitch" and some pencil head is sitting in Barbados with your hard earned retirement.
If no one knows, then it's really no biggie. But it is just a simple question. If you can't contribute to a useful answer, then what's the point of posting?😲
Originally posted by nilocto make a long story short, you are calculating it wrong.
I've noticed a slight calculation error when tracking my own rating. The difference between my rating calculation and RHP's is only one point (at this stage).
Is a person's rating truncated to be a whole number? That, which is displayed. Or, are there hidden decimals that we don't ever see, and that get included in one's rating, and calculated against an opponent?
Kind of like that 1/2 penny in Jumping Jack Flash, with Whoopie.
the way thay handle the inner representation of the formula doesn't matter. it wouldn't matter even if they calculated the ratings by having a blind, drunk ape throwing darts, because it would still be The way the ratings are calculated.
Originally posted by nilocI'm not sure, but I would expect that RHP does truncate a person's rating to a whole number. The reason for my assumption is that the USCF calculations also truncate rating numbers.
I've noticed a slight calculation error when tracking my own rating. The difference between my rating calculation and RHP's is only one point (at this stage).
Is a person's rating truncated to be a whole number? That, which is displayed. Or, are there hidden decimals that we don't ever see, and that get included in one's rating, and calculated against an opponent?
Kind of like that 1/2 penny in Jumping Jack Flash, with Whoopie.
Check out the last paragraph on Page 10 of Mark Glickman's paper on the USCF rating system -
http://math.bu.edu/people/mg/ratings/rating.system.pdf
Edit - Actually, truncation isn't the word I should have used. The USCF calculations either round the rating up or down to an integer value, depending on certain conditions.
ratings are calculated to four "hidden" decimal places. however, the RHP accountant is "skimming" the points and depositing them with a secret player who has one of the highest ratings in RHP, even though the player rarely wins a game against anyone stronger than 850.
Several members have complained about the "missing" or "stolen" hundredths of a rating point, but alas, as it turns out, it is quite difficult to prove the illegal and unethical "transfer" of unearned fractions of rating points.
The question is open: where are all those fractional rating points going? Are we all being denied our "true" rating? This is an outrage.
Please join the RHP members coalition to fight the loss of hidden rating points (RHPMCFLHRP)!
I belive the concept of 'skimming' points is more then a mere fabrication. I have frequently played a series of blitz games against someone of a similar standard and we score 50% in the matches but end up with both our ratings going down.
I am sure there is some greedy fat cat somewhere who is hoarding all those points.
Originally posted by TyrannosauruschexCurses, they're onto me.
I belive the concept of 'skimming' points is more then a mere fabrication. I have frequently played a series of blitz games against someone of a similar standard and we score 50% in the matches but end up with both our ratings going down.
I am sure there is some greedy fat cat somewhere who is hoarding all those points.
-ducks-
