chess.com ratings are deflated against USCF

I don't think a math degree has anything to do with the issue, nor the number of math or chess books you've read. Nor any chess title whatsoever.

I think you can compare, no doubt, but just as polling is a science, you have to be careful of the noise. If you could somehow take all the non-cheating chess.com players which have enough games played and do not have stale ratings, then you'd have what you want I think. I don't know how close it would be to comparing just the publicly-available averages. Myself, I usually hover in the low-2200s on chess.com blitz, which is within 25-50 points of my USCF rating.

Wow, I cannot believe what I am reading.

First off, taking the average of the overall USCF rating and subtracting the average blitz rating from it may be completely meaningless. Ok, you are trying to make a very simple linear best fit line (simple linear regression). Why did you choose linear regression over, say,  polynomial regression? What about logistic regression (Ok, I can't remember if this actually applies here)? Some nonlinear regression?  What is the correlation coefficient and what are the ranges of your confidence band (say, 95%) for the corresponding variables?

Seriously. Trying to apply a very basic model to what might actually need a more complex fit is a little silly. I'm curious as to how you came across to your conclusion (why linear regression over potentially better fitting regressions?), how many data points did you use (is the amount used significant???), and how confident are you in the best fit line that you chose?

I understand Adam continuously harps that "an average is good enough." However, the question has to be asked though: why is it good enough? Or is it simple the lazy approach?

keep it simple silly #butnottoosimple

Francisco, it would be interesting to see a statistical analysis of the OP's thesis performed. The OP has produced a hypothesis and declared it to be the truth. It seems that "North Korea" is just as bad at producing pseudo-scientific gobbledygook as some American high schools. (It's not the fault of the schools, which are doing everything they can to fight the anti-scientific attitudes of many communities).

Anyway, as I said, it would be interesting to see what sort of correlation and how strong such a correlation is, between USCF and chess.com ratings.

I'm not holding my breath to see if anyone rises to that challenge tho.

I corrected it and then you made it wrong again.

Hey! nobody corrected my Gran'ma'!  She'll show you the back of her hand.

I really want to know how you got your sample.  I attempted to get my own sample and ran into problems.  

First, I searched the member list for "David" and sorted by most recently online - this ensured that the people had sumbitted something like an actual name in the Name field, and that the person was an active member of some sort.

Going through the list, I skipped anyone who had a location outside the US, or a location of only US (I needed the state in order to be reasonably certain they were the same person on the USCF list.)  For those who met that criteria, I clicked on their profile.  From the profile, I checked to make sure they had a first and last name listed (many had just "David" or "David S" and could not be used.)  Then, I checked if they had a blitz rating, and if they did, I made sure it had enough games to be reasonably accurate (I used >20 games.)  If they met that criteria, I then typed their name into the USCF website.  I threw out anyone with an old or provisional rating (although most were not listed at all.)  I also had to throw out a couple of people with ultra-common names because there were multiple people with the same name from the same state.  I did not encounter myself but would have thrown that out as well.

After going through 30 pages of results, I ended up with a grand total of six acceptable results.

For what it's worth, the USCF ratings of those few good samples were 1986, 1369, 1923, 2058, 1272, 1808.  The chess.com blitz ratings were 1709, 1245, 1581, 1943, 1278, 1554.  On average, the difference was 184.33.  The standard deviation of the ratings difference was 128.8.  In conclusion, I cannot tell whether 250 is close to accurate or not.

Of course, you have a problem there when you accept that your source data is valid.

Chess.com claims an average rating of all members is 1140, but when you look at the average rating by country, you'll notice that all but a few small nations have average ratings a fair bit higher. Shouldn't the overall average be close to the average of the most populous nations

Having just checked the ratings by nation, I can verify that the only nations in the world that are below the world average are...........

South Georgia, Chad, Falkland Islands and Grenada.

So if you happen to live in Los Malvinas.....try harder

And the USA average rating is 1275, so your deflation is only about 100pts, if you assume that most USCF members are from the USA

A simple average is good enough. If you want more, do it yourself. Jesus. You talk about being lazy? So do it yourself, man. Its like you people just want to say the stupidest things. Am I getting paid to do research on chess ratings? Because, if I'm not getting paid, you are gonna have to be happy with a simple average.

Um, well, 184 is getting awful close to 250 -- don't you think? Since I originally posted 2-300, and 184 rounded to the nearest hundred is definitely 200... it sounds to me like you just independently verified what I said.

I was making a point. The point is that a math degree has nothing to do with this. This is not that complicated. You don't have to be a genius to figure out that USCF standard ratings tend to be several hundred points higher than chess.com blitz ratings. Why this is controversial is beyond me. I think some of you just want to argue.

I don't think you have a maths degee, Adam. Surely anyone with a maths degree wouldn't say anything as presicely meaningless as 'post-calculus'.

Nonetheless, I think you're probably right. There likely are relationships between chess.com ratings and USCF, but I shouldn't think any relationship is likely to remain static as the userbase of each changes. 

"I think some of you just want to argue." 

Just by reading your first few posts, it's evident you are perfectly happy to enter into good old fashioned, pointless arguments.... in fact, your very first response to a perfectly pleasant post includes an accusation that the poster can't properly read.

70 point different 180-250 is fairly large.  That's like a third of a class.  Plus look at his standard deviation and sample size.  Clearly a "Math Doctor" like you should understand that drawing a conclusion from that is unreasonable. The fact is you can't correlate blitz ratings to over the board uscf.  Some people will have higher blitz than otb and some lower.  Some will have blitz ratings close to their uscf.  But the point is the population you are using has lower blitz ratings compared to their uscf and your sample size is too small, and not random which means it violates basic statistical rules to even remotely make your conclusion valid.   The only reason you want to believe your results is because you are trying to JUSTIFY your own poor blitz rating compared to your uscf.  

<CASE CLOSED>

Yeah I think he threw the degree in there for good measure when people weren't agreeing with him quick enough. The 'post-calculus' remarks sounds suspiciously like something someone who's starting to learn calculus might say.

I'll try one more time, but don't have much hope.

Please provide the information on the approximately 20 people you know, (or two dozen, or dozens of players) so that someone else can at least pretend the sample is random and representative of the group of players who have both USCF and chess.com blitz ratings. This will allow the rest of us to see how you reached your conclusion and why you are so convinced of its absolute certainty.

Oh, no, see, this is why I can't take most of you seriously, because you say absurd things like this. Its not uncommon for someone to swing 70 points in a day, and the guy who came up with the figure of 180 stated himself that he used a much smaller sample size than me. However, since 180 is clearly quite close to 200, I think his numbers do indicate that there is a significant drop from USCF standard to chess.com blitz.

Can you people just accept that? If you want to average my results with his, ok... the average difference is 215. That sounds good enough to me. If you want a more accurate result, for reasons which elude me, you will have to do the research for yourself. I think we've clearly established there is an average drop of 2-300 points between USCF and chess.com.

Guys, you can laugh about post-calculus if you want, but all it means is I took calculus, and I kept going. Most of you probably haven't even started calculus, and very few people go past a first or second course. However, I didn't need to use any of that in order to calculate a simple average. I don't need to employ statistics, in order to understand an obvious trend. Watching cars driving down the highway, I can quickly deduce an average velocity, without any complex equations. I guess maybe some of you don't know how to do that?

Looking at some of these posts makes me think that people are mixing apples and oranges. 

My USCF correspondence rating is 1995 and my ICCF correspondence rating is 2087 but I have not played correspondence chess with them in a very long time.

The only correspondence type chess I play is here on chess.com.  In the USCF I play in long OTB tournaments which there is no comparison here.  Blitz chess I do not play.  In fact in my opinion it is more "beat the clock" then it is chess.

My USCF OTB rating is 1678.  Correspondence type venues will naturally be higher.  On chess.com as it is in the USCF or the ICCF you can use opening books, databases etc etc.  In OTB you are on your own.  Nothing can be used. Also the thinking time is much longer in Correspondence style chess.  Therefore a well played, well thought out correspondence game should approach perfection even though the majority do not but blunders are rare and a one pawn advantage many times means a won game.  In OTB chess there are far more blunders and mistakes. 

In Blitz chess the time factor plays a very big role more so than actually knowing the game well.  I refuse to play blitz because I have recently trained my self to examine the board and take my time in calculating a response.  If I play blitz I get back in that "move fast" mode which I do not want to do.  In my area I am friends with a USCF Master that goes to tournaments and plays in the Blitz games held a day or so before the long games start.  He is very good in both areas.  I am not that fortunate or that good!! 

I use chess.com to practice for OTB therefore I have not played many "live" games here.  I use the tactics trainer, chess mentor and watch videos.  I also use the "analyze game" feature which points out trends that can be corrected.  Would not know if I did not have the game analyzed.  I also recently purchased a chess coach form the list here and he has helped me immensely. 

Chess.com is a great site for learning and improving.

Chuck

Yah, I used to be that way, however my friend is 2150 (USCF) and 1900 (chess.com) -- notice the 250 point difference? He encouraged me to play more blitz, saying that if you can still maintain the discipline to think during longer time controls, your blitz experience will make you more efficient.

I wasn't quite convinced, but then I heard Dan Heisman say essentially the same thing. He said that beginners should not play blitz, because they won't learn to think. However, after one learns to play longer time controls, he suggested that blitz practice does gain value. However, this is only true if the blitz player takes time to analyze the games afterwards.

This is why I would guess that although an experienced standard player will experience a 250 point drop from USCF(standard) to Chess.com(blitz), the opposite is not true: an experienced blitz player cannot expect a 250 point increase when switching to standard, as they probably lack the mental discipline. However, if they manage to learn how to manage their time and concentrate at longer time controls, they probably will experience a 250 point increase.

Not talking to you RunawayKnight:

Its a shame that the trolls just come here to argue, spamming and preventing us from actually having a discussion. I really wish people would realize: if you don't agree with an OP, just get lost. Start your own thread. If the OP wanted your opinion, he would have asked for it. Arguing with someone who started a thread is about as classy as interrupting a professor, minister, teacher, or anyone giving a speech. If you don't agree, just walk away.

Why do some people have this obsessive-compulsive need to fight? Is it because they have low self-esteem, so they come to the internet and try to quibble? Seriously, get a life people! Do you notice how I never come to your threads, and try to argue with you? Its because I don't care what you think! Take a hint. I'm only interested in people who are smart enough to understand what I'm saying. If you don't get it, I have no desire to explain it to you. I certainly am not interested in arguing with a pack of internet morons.

"Watching cars driving down the highway, I can quickly deduce an average velocity, without any complex equations. I guess maybe some of you don't know how to do that?"

What an odd thing to say. Is this supposed to prove to us that you're an accomplished mathematician? ... so you can time a car over a roughly estimated distance and employ that most confounding of equations: V(av) = delta(s)/delta(t) ?  

You're a high school student. 

"It's a shame that the trolls just come here to argue, spamming and preventing us from actually having a discussion."

You don't come across as so noble when you have to lie about your background to assert your case.

Hi Adam as you say

"Yah, I used to be that way, however my friend is 2150 (USCF) and 1900 (chess.com) -- notice the 250 point difference? He encouraged me to play more blitz, saying that if you can still maintain the discipline to think during longer time controls, your blitz experience will make you more efficient."

---------------------------------------------------------------------------------

I am 70 yo and I will not live long enough to get that discipline.  I find it very difficult to think fast and move fast.

At a very recent local tournament I was playing a 2100+ player and had him beat.  I had 10 min. left on my clock and he had 7 sec.  But we were playing a 5 sec delay.  He made 10 moves in 3 sec off the clock with 4 sec left on his clock I resigned.

The problem was that when he started moving at ultra blitz speed I tried to match him but I had the time on my clock.  Why did I start moving as fast as him?  No need to do that.  If I would have taken my time and played he did not stand a chance.  He recognized that and told me so after the game. 

For me thinking fast and moving fast is poison to my game.

Chuck