chess.com ratings vs USCF

Is there any objective data comparing USCF ratings with chess.com ratings?  It is my suspicion that most internet sites which give ratings have ratings which are dramatically inflated relative to USCF.  For example, in our club I know someone whose USCF rating is around 1400 who brags that he has an 1800 internet rating on another website.  But the few people I know on chess.com have ratings here that seem quite close to their USCF ratings.

Is there any data on this?  If not, can I at least get some feedback from other users about how their USCF ratings compare with their ratings here? 

I played a 1616 once who claimed a USCF rating of 1280.  

I have seen a 2800+ who claimed a USCF in the 1900's.  

That's 2 examples right there. I can be a third example after I play in my first otb tourney but that will be in a few months.

I don't think anyone's done an exhaustive statistical study. But even if they did, the comparison is difficult because it depends on how seriously you take the games.

On the ICC, I have a rating that is thought to correspond to about 1700 USCF. But I take those games very seriously, frequently more seriously than my opponents. I routinely beat people who aren't putting their best effort into the game. Heck, I once drew an IM for that exact reason. Those same people would have no problem wiping me off the board in an OTB tournament game, where they play with their full effort.

On chess.com, I play to goof around. Just based on the guesses people make, my rating here might correspond to 1100-1300 USCF. I play unsound moves just to see what happens, I play correspondence moves almost instantly with no analysis. I routinely lose to people that I'm pretty sure I'd have no problem with in a serious tournament environment.

That's why OTB tournament ratings are the gold standard. Everything else is just goofing around.

It's definitely the case, though you should also note that the type of competition is completely different. USCF is usually slower time controls, it's OTB, it's a tournament setting, you don't typically choose your own opponent. Live chess here is usually blitz with self-selected opponents. Online (turn-based) chess is a completely different beast from both. Some people play it very casually, some play it the same way they would a "serious" correspondence game, some play it somewhere in between the two. Suppose I am "objectively" a 1600 player and I am playing against another "objectively" 1600 player in turn-based chess, except I spend only a couple minutes, maybe a little more, on the position when it's my turn and my opponent treats it like a serious correspondence game. He will almost always beat me. If we carry out this philosophy to the extreme, he will have a far higher rating on chess.com. However, if we face each other in an OTB game at a USCF tournament, we'll probably be 50-50 and we'll perform at around 1600 level OTB. It's a completely different game.

And note that rating isn't an objective measure of how well you play chess, it's a measure of how well you perform against other players with ratings. As such, you can only really compare internally - there are no real "benchmarks".

Many members do post their USCF or FIDE ratings on their page...

I wonder if it is a simple thing for chess.com to simply list the collective ratings in a table, 1 column for chess.com rating, 1 column for FIDE rating, 1 column for USCF rating (or whatever national rating system), etc., and it would obviously change as people's ratings change and they update their other ratings.  And also list the number of members that are used in the calculation...that should be relatively simple I'd believe and a very interesting exercise.

My chess.com rating is uncannily close to the FIDE equivalent of my last BCF rating nearly 20 years ago. But I am just one data point.

I do not always take my online games seriously - but perhaps neither do my opponents. Perhaps it evens out.

You can compare internally in both rating systems, and if you fall within a similar percentile in both then you would be able draw conclusions from the relationship you find about the group you are in.

This is my standard chess rating and percentile for chess.com

This is my blitz rating and percentile for chess.com

I have not played enough OTB tournaments to make the following correlation but I found a graph at blogspot.com on fpawns blog that will illustrate a relationship

  Graph by fpawn @ fpawn chess blog

Now, if 1632 = 95-97th percentile chess.com scale
and roughly 1900 is the 95th percentile for USCF in northern california then we have a relationship to look at.

conclusion: Either 1632 chess.com = 1900 uscf (my guess is that is highly unlikely) or the competition on chess.com is lower quality than that of USCF events in northern california (seems more plausable).  With easier competition a higher rating is possible with relatively less chess ability therefore I would assume that i have an inflated percentile/rating relative to the percentile/rating of a uscf rated player in northern california.  If we assume that california uscf trends the same way as the rest of uscf ratings then it seems like it would be possible to build benchmarks if we get enough data points from players who have both uscf and chess.com ratings.  This would create a relationship to eliminate the variable of serparate talent pools and make comparisons and benchmarks possible to predict.

I would be interested in looking into this.  does anyone know of a resource where I could find a larger uscf rating sample?  Any ideas for the best way to gather uscf and chess.com ratings?

Not to quibble, but it looks like 95% is closer to USCF rating of 2000 rather than 1900.

I think the correlation is correct since at my 1374 rating (68.5th percentile), that would put me at just under 1500 USCF, which is accurate considering I have never played in an OTB tournament ever.

And monkeys fly out of my butt

There are about 15 times as many chess.com accounts as active USCF players (active USCF means played a rated game in the last year). chess.com accounts are also strongly biased toward the low end of skill because of the low barrier to entry. A lot of people barely know the rules! The top end is of ability is similarly biased against because there's no reward for them to play here - there aren't enough strong players to play against and they can compete elsewhere. For the most part. There are some exceptions. Also, you start at 1200 here. So that makes 1200-ish a sort of default rating for the "ignorant masses".

we ran numbers here that compared people's "stated" uscf/fide rating with their Chess.com ratings. we were slighly overvalued by about 50 on echess (turn-based), 100-200 on live, and about 100 on tactics, and about 300 on chess mentor. we've put some things in place to bring those numbers more in line. but there are all sorts of reasons why numbers can vary. some people play in a tournamnet once a year (like me), but play a lot online. 

Thanks to all for the info.  It seems just as I suspected that the ratings are inflated by about 200 points.  Sigh.  I thought I'd improved 200 points since I got here.

NrthrnKnght I can sympathize.  The expense of competing in regular OTB tournaments prices most ordinary family guys out of the market.  One alternative however is to do it yourself.  Become a TD and organize your own tournaments locally.  I do this myself.  Our local club meets at the Barnes and Noble.  I run four quarterly Swiss tournaments G/75 one game per month.  I charge just $5.  Nobody has to pay for overnight accomodations or meals.  It's cheap and it's real chess.  Not as classy as a big tournament in a hotel but hey it's affordable.

oh

If you don't take online game seriously as tournament games, then you don't enjoy chess, you only enjoy tournaments prizes.

Not true. I enjoy chess, but I still don't put as much effort into my online games as in the tournament games. Why would I think about my move for 3-5 minutes online? I see those online games as practice, where I can try out stuff that I wouldn't risk in a real game. Also, I play quite a lot of games at the same time online, while I obviously only play one game at the same time in real life.

I think I have to disagree with the consensus here. I've observed that many NMs and FMs are rated 1900-2000 for blitz on chess.com. I would make an educated guess that the majority of NMs and FMs are rated below 2200 on chess.com blitz. It would be interesting to poll certain USCF classes (class A, expert, master) on the difference between their chess.com and USCF ratings. It may be that there's a cut-off point where it becomes harder to get points on chess.com than it is in USCF tournaments. 

I'd be interested in someone proving me wrong, if that's the case. I really am curious what the average chess.com rating is for NMs and FMs.

     Doesn't Chess.com inform players how to convert a Chess.com online-chess rating to a USCF rating?

Don't be fooled by these numbers. They only go one way.

Because someone has a USCF rating of 1800 and a Chess.com rating of 1600, does not mean you are the equivelant of an USCF 1800 player because you have a 1600 rating on Chess.com.

ANd, as mentioned in post #11, there are way too many variables for any of these numbers to be accurate.

http://www.chess.com/article/view/chesscom-rating-comparisons

They are not inflated they are deflated. Most 2100s on chess.com play around 2300. Maybe not so for lower rateds but definitely for higher ratedss