chess.com ratings are deflated against USCF

It's just a fact. Other countries are better at chess. Deal with it. Americans in the USCF can expect a ratings drop when they are put into a global pool. The same thing happens when they join FIDE.

For the record, you need to work on your reading comprehension before you chime in. We aren't talking about the chess.com "online rating". If you can't figure that out after 37 pages, you aren't trying to think. Furthermore, you don't have a standard USCF rating. Stop wasting our time pretending like you are part of the conversation. Just go away, this thread doesn't need more spam from people who can't understand what we are talking about.

His opinion means more than yours. At least the guy in papua guinea understands what I'm talking about. Maybe you should stfu so you can learn more about the real world?

Anyone who says someone from new guinea can't have a valid opinion about chess, and needs to move to America in order to understand reality... well, anyone who talks like that is not someone I can respect.You should be ashamed of yourself.

I just read the first two pages.

Of course there will be an average rating difference between any two chess rating pools, regardless of the kind of chess game played. Chess is chess, and if you're good at one particular kind of chess you will probably also be quite good at other variants of the game. There will always be a correlation, but it might be weaker or stronger depending on the types of chess. The correlation between USCF and FIDE is probably quite high, and the correlation between USCF/FIDE and chess.com online rating is probably much lower as USCF and FIDE have much more in common.

So if we compare any two variants of chess, there will always be a calculable mean difference and correlation coefficient, and a 95% confidence interval for the mean difference should be easy to calculate. Of course, there are many potential sources of error as well as measurement difficulties.

Are the estimations that AdamRinkleff (AR) give correct? To assess this we have to address the potential sources of error. A few key questions that come to my mind are:

1. Is the sample that AR uses large enough? - I'd guess it's large enough to detect that there is a difference, and the direction of the difference, but I think the confidence intervals would be very wide.

2. Is the sample that AR uses representative for the entire population of players with both a USCF and a chess.com blitz rating? Probably not, and this will introduce potentially heavy bias in the results, which is reason enough to disregard the results from a scientific wievpoint.

In conclusion, I think that the main point of AR may be correct - that there is a measurable difference in ratings between USFC and chess.com blitz, but I think his data are too weak and biased to draw the conclusions that he does.

Let me put it this way, if I was offered a million dollars for winning a chess game, and I had a choice:

1) 5 minute blitz on chess.com versus a 2000
2) 90'30 standard in USCF versus a 2000

I'd play the USCF game. I've beaten USCF 2000s in standard, but never a chess.com 1800 in blitz. Indeed, I know several USCF 2200s with a chess.com blitz rating below 2000. Their skill is approximately the same in both pools, but their rating is much lower on chess.com. This difference is pretty consistent for all players who have ratings in both systems. You do the math.

That's all I've been saying. I estimate the difference is about 250 points, you can calculuate it yourself if you want more precision.

Send me the data and I'll calculate it. But the main problem is the lack of representativeness of your sample for the whole group so we can't really draw any conclusions from it regardless of what comes out of the statistics program.

I do think that you're right though, that chess.com blitz ratings generally are a bit lower than USFC ratings in people who have both ratings. Anyone who talks about apples and oranges need to learn statistics or at least realize that they don't know what they're talking about.

Well, the numbers are in.  It looks like for most chess.com blitz ratings of 1500+, the difference from their USCF rating is <= 100 points.  And at a quick glance, I don't see any more than an ~ 150 point difference at the low end.

So much for that 200-250 point difference.

http://www.chess.com/article/view/chesscom-rating-comparisons?page=1

In my measurements, I only included people who were active in both pools.

The data provided by chess.com is based on users submitting their own numbers, which is not reliable. As seen within this thread, a lot of people submit their USCF rating when they haven't played in recent tournaments. Consequently, since they have continued playing chess here, their USCF rating underestimates their current skill. Most of those people would have a rating increase if they played more tournaments.

If chess.com wants an accurate measurement, they need to collect people's USCF member ID, so the data can be verified.

I havent followed this thread but I agree 100% with this.

I just had a draw a couple weeks ago OTB against a 2000. And I had a easily won game at one point and missed the two move tactic (oddly enough it was a tactic I had built the position for and just overlooked it)

But blitz against a 2000 here? My chances, while not exactly zero, are close enough to zero to make it barely worth trying for.

While it's possible that chess.com members engaged in a massive disinformation campaign in order to discredit a 15 month old post they didn't know about, the compiler asked that no one submit a rating for any category that didn't contain at least 25+ games with 5+ of the games being within the last 6 months. So yes, he was looking for players that were active in any pool they submitted a rating for.

Yes, it's likely some of the data didn't follow that criteria, but to try to throw out everything because it disagrees with your 200-250 point rating hypothesis is amusing.

Please don't make exaggerated 'strawman' arguments.

I never said chess.com members engaged in a "massive campaign of disinformation". I did say that you can't trust self-reporting. Take a basic class on statistics or psychology. I repeatedly stated in this thread that I was only interested in people who were active in both pools, and I had responses from numerous people who had not played in USCF for a year or more. One guy didn't even have a USCF standard rating, although he genuinely believed that he did (he had a provisional 'quick' rating). I'm sure those same people submitted their data to chess.com, and I doubt most people read or heeded the requirements of that survey. Furthermore, the chess.com criteria was very weak, five games in the last 6 months, which isn't exactly active. The people I was looking at were playing tournaments every week, sometimes more than one per week.

Its really annoying how this thread has shifted from, "You are an idiot because there is no way chess.com ratings are deflated." to "You are an idiot because chess.com ratings are only deflated by 100 points." You obviously just want to disagree for the sake of being obnoxious.

Once again, repeating myself ad nauseum, if chess.com wants accurate results they need to collect the USCF ID numbers so we can verify that the people who submitted data are actually active in both pools.

According to my colleague, CC player, and former TD, most CC ratings for USCF members are 200-300 points higher than their OTB ratings.  According to him, it "goes with the territory."

Just look (informally) at ratings on Chess.com.  Essentially the same pattern holds for Blitz ratings.

Yes they appear (informally) as deflated compared to USCF OTB ratings.

And both conditions appear to have held for the past 15 months of this crazy thread. Get over it.

CC has nothing to do with this discussion.

Inflation/deflation can only occur within a single rating pool.  For example FIDE ratings becoming inflated - ie. going up on average over time disproportionate to increases in playing strength.

A rating cannot be inflated/deflated compared to another type of rating, a rating only indicates your strength relative to the other players with the same type of rating.  It does not indicate absolute playing stength or have any obvious comparison with a rating from a different pool, although it may be possible to correlate ratings from different organisations to see how they compare.

Inflation/deflation are not terms relevant to this discussion.

Haha, fair enough.  Has there ever actually been a study to try and determine with any degree of accuracy the relationship between ratings from different pools or is it all just speculation?

USCF scholastic ratings are grossly deflated.  Letting kids start at 100-900 ratings and rise @1000 points in 2 years (or less) is just plain bullshit.

All it guarantees is that older players (in the D, C, and B Classes) will lose games to these kids, get demoralized, and drop out of the tourney scene. To no one's benefit.

No one should start with a rating lower than 1000, E Class.  Very simple.

Mr. Glicko is trying (vainly) to rectify this situation of rapidly rising kids effects within the rating pool.  Glicko-2 is his attempt to rectify this, along with dozens of other statistical concerns that are bogged down in USCF discussions.

But the USCF remains litigious.  So progress is very slow.  Kinda like progress in this thread.  15 months and counting.

P.S. the habitual and chronically rote comments about "rating pools" (as per above) are also just plain bullshit too.

But whose listening ??  No one it seems.

So Have Nice Day.

I never said that you said that, just making fun of your flawed premise.  Kind of surprise you haven't slinked off to another corner of the internet now.  Yes, they all magically happened to wrongly self-report in a way that made your hypothesis wrong.

I agree with this about these kids. I almost got beaten by one of them at chess club. He was rated 600ish. He was every bit as good as the 1400ish players I was posting a .500 record against. I actually question if this is done to encourage these kids, by tricking them into thinking they are better than they are, so that they dedicate themselves to chess ?

How so?