chess.com ratings are deflated against USCF

Sort:
PeterB1517

This discussion isn't on cheating... but really, why would people be cheating.  We're not playing for money.  In limited time controls, it is difficult.  It is hard to use another computer program with chess.com.  In long time controls, it is very possible, but you get back to that motivation factor again.  Mostly, people are on here because they like to play the game, and think if they play more, they will get better, which sadly seems to be untrue.

nameno1had
Ubik42 wrote:
nameno1had wrote:
bigpoison wrote:
zborg wrote:
bigpoison wrote:

You're obsessed dude. Cheaters aren't playing bums like us often enough to notice. The ops reckoning ain't dead, but it's close enough.

Cheats are playing at all rating levels.  Intermittent cheating, is quite common.  They use their engines during small parts of the game.  Often times this is enough for them to win.  Sometimes not.

And when they "turn on" their engines, they (sometimes) start to taunt you in the chat box.  It's all very entertaining.  Except then, you typically lose.  

I've been playing chess here for six years.  I've played four opponents who were later banned for cheating.

If there are folks who use an engine for only a couple of moves a game, I'd never know the difference anyway.  Why do you think there are many folks who do this?  What's the point?

I see many who intentionally play worse moves than the best, that were obvious, to lower their t-stat scores that are used in cheat detection analysis. People here have some real misconceptions about what it is like to play against someone, who using engine assistance in various forms.

Of course people cheat here, but this gets a little too machiavellian. Why on earth would you combine cheating with intentional bad moves? Does this make sense to anyone? You play a flawless computer move, then balance it out with a blunder? What is the goal?

bigpoison

What the hell is an obvious best move anyway?  Is that a forced move?  If it's not, it's obvious to whom?  Obviously, what's obvious to trysts isn't obvious to me.

Some people get a decisive advantage using an engine and then play the least highest scoring move according to analysis, so that the resulting mass stats from the games don't reveal anything fishy. This usually isn't discovered until the games are looked at individually and according to the deviation scores.

Some people think things are always so simple, until the are more closely scrutinized....

zborg
bigpoison wrote:
I've been playing chess here for six years.  I've played four opponents who were later banned for cheating.

If there are folks who use an engine for only a couple of moves a game, I'd never know the difference anyway.  Why do you think there are many folks who do this?  What's the point?

They like to win, and (intermittent cheats) won't get caught.

Just like the people who play their first 20 moves out of the MCO sitting in their laps.

Both are VERY SIMPLE techniques, don't you think ??

zborg
chess_gg wrote:

Chess players are nutsy. That's why.

+10, And it's proven daily in these forum discussions.

This one is fully 15 months old, and still going strong.  Great Job @AdamR.  Laughing

zborg
The1899Club wrote:

Why do you insist in hijacking the discussion with your rantings about cheating?  You are merely attempting to draw attention away from the fact that you are too dull to understand how ratings can be compared.

And you live in a pinheaded world of hairsplitting exceptions.  So tiresome.

Your concept of rhetorical argument is to grab onto a single exception and PULL HARD, fully expecting that the knitted sweater will unravel.

Sorry.  It doesn't.

 

DrCheckevertim

Cheating is actually relevant to the topic. If 10% of people cheated on chess.com (just as an example), that is a strong factor in rating pool deflation.

bigpoison
zborg wrote:
bigpoison wrote:
I've been playing chess here for six years.  I've played four opponents who were later banned for cheating.

If there are folks who use an engine for only a couple of moves a game, I'd never know the difference anyway.  Why do you think there are many folks who do this?  What's the point?

They like to win, and (intermittent cheats) won't get caught.

Just like the people who play their first 20 moves out of the MCO sitting in their laps.

Both are VERY SIMPLE techniques, don't you think ??

Yes, simple techniques, but if the point is to win, how can there be engine users at all ratings levels?

Why would someone cheat occasionally to maintain a 1700 chess.com rating.  It doesn't make any sense.

soupram

mmmm...

DrCheckevertim

I could see how someone would cheat occasionally in order to win more than they usually would, and yet, avoid getting caught or bringing up suspicion. That's just a smart cheater. It surely happens less often though. I don't think most cheaters even care about covering their tracks.

AndyRichter

As a gay man, I take positive representations where I can get them. Any time a same-gender relationship is portrayed in a positive but very real light benefits us all. The same can be said of Chess.com's anti-cheating efforts, which, much like being gay, will likely remain unpopular in a world that seems married to FIDE ratings, and never really "come out of the closet" and be truly ready for acceptance of the online chess rating. But anytime we can get some good press, it helps us all. I'm a big fan of Chess.com (even over FICS!) and I'm proud that Chess.com has taken a stand and acknowledged that some of us are different, and that's ok.

zborg
The1899Club wrote:

I bet your pardon.  I directed my comment to the other fellow.

Mea cupla.  I apologize, assuming you directed that comment at me.

But who knows ??

Oecleus
bigpoison wrote:
zborg wrote:
bigpoison wrote:
I've been playing chess here for six years.  I've played four opponents who were later banned for cheating.

If there are folks who use an engine for only a couple of moves a game, I'd never know the difference anyway.  Why do you think there are many folks who do this?  What's the point?

They like to win, and (intermittent cheats) won't get caught.

Just like the people who play their first 20 moves out of the MCO sitting in their laps.

Both are VERY SIMPLE techniques, don't you think ??

Yes, simple techniques, but if the point is to win, how can there be engine users at all ratings levels?

Why would someone cheat occasionally to maintain a 1700 chess.com rating.  It doesn't make any sense.

Perhaps they are 1500 and are trying to delude themselves into thinking they are better by only using a few computer moves. By the way, this whole "only use a few moves from the computer" is probably relatively common. I used it to cheat on fics (I've never cheated on chess.com). And fics never banned me. I would typically use it when someone was talking trash and you wanted to beat them, or just for fun, or if you are in a bad mood. I don't think that way anymore and I dont ever cheat, just letting you know why someone would.

zborg
bigpoison wrote:
 

Yes, simple techniques, but IF the point is to win, THEN how can there be engine users at all ratings levels?

Why would someone cheat occasionally to maintain a 1700 chess.com rating.  It doesn't make any sense.

If you are here to improve your playing strength, win games, and avoid detection, it makes perfect sense to use engines intermittently, get a 1.5 pawn advantage during some part of the game and boost your rating while still making the lion's share of your moves.

You typically end up with a rating roughly 200-300 points higher than you would have otherwise.  And you keep practicing and improving along the way.  Except you're using a crutch to win games, more often.

Only a "monkey" lets the engine make all the moves, and wins nearly all the time.  Yes, those ratings go sky high.  But that Monkey eventually gets caught, and they learn essentially nothing along the way.

Why does this sound unreasonable to you ??

Let me suggest that your "IF / THEN" statements (as per the quote above) is begging the question.  The answer you support is already embedded in manner in which you poise the question.

God, what I just said was boring syllogistic crap, wasn't it ?  Laughing

On a side note, it's too bad the Steelers didn't make the football playoffs this year. Even they would have played better than the Broncos in the "Soup Or Bowl," as per the Jon Stewart show last week.

Irontiger
The1899Club wrote:

I wonder why the mere thought of comparing federation ratings to internet ratings gets some people so worked up(...)

Hardly a new thing in this thread. The OP managed to start the insults as soon as post #3.

robthepek

I agree with the OP. My Uscf is 1500-1600ish (1698 currently), and my chess.com blitz is about 1400.

SmyslovFan
SmyslovFan wrote:
PeterB1517 wrote:

Petrip wrote:

This is the best analysis of the thing

http://www.chess.com/forum/view/general/fide-ratings-vs-chesscom-ratings-explored

Important point is that correlation coefficient is merely 0.738 for blitz rating-Elo. Meaning that only about 50% of variation in chess.com rating is explained by ones Elo rating or  the other way round. Rest is explained by differences of the two different environments.

and here is the Elo prediction formula from blitz rating, which seems to be the most reliable predictor:

483 + 0.769 * Blitz = FIDE rating (+-193)

So an error of 200 points. So you  compare, as long as you undersstand that result might be off by 200 points and is unlikely to hit the target within 50 pts.

And chess.con Std to Elo has corr. coef. below 0.6 meaning that only 36% is explained by rating in one pool and rest is something else

 

It was a good link and it supports Adams original point that chess.com blitz ratings are lower than OTB rating by about 150 pts. [Added: Give or take 200 rating points]

So, if you're satisfied with being accurate to about one standard deviation, you can say there is some correlation. Yes, ratings can be compared, but a person cannot accurately predict within 50 rating points what their FIDE rating will be based on their chess.com rating. They could be off by 200 rating points in either direction.

I don't like quoting myself, but I think this gets to the heart of the matter. 

Who here believes so strongly in Adam's point that they would go to a FIDE tournament and tell the TD they should play in the 2000-2200section because the formula at chess.com suggests they are 2100 strength?

The margin of error is a full standard deviation! That's not a reliable difference, and for someone who claims to know math to compare it to converting mph to kph? Really?

ipcress12

SmyslovFan: If you read Glickman's paper on his rating system ( (http://www.glicko.net/glicko/glicko2.pdf ) you discover how loose ratings are within their own universe, much less being compared to other rating systems or time controls.

As with the original Glicko system, it is usually informative to summarize a player's strength in the form of an interval (rather than merely report a rating). One way to do this is to report a 95% condence interval. The lowest value in the interval is the player's rating minus twice the RD, and the highest value is the player's rating plus twice the RD. So, for example, if a player's rating is 1850 and the RD is 50, the interval would go from 1750 to 1950. We would then say that we're 95% condent that the player's actual strength is between 1750 and 1950.

I've implemented Glickman's system and run it over a population of chess.com players and discovered the RD starts at 400 and takes a good while to reach 50 or less.

Until a player has played 50-100 games or more, depending on tau and the rating period, the RD is more like 200 in which case the rating is +/- 400.

So there's no point in getting huffy about errors of +/- 50-200 points.

Ratings aren't meaningless but they aren't nearly so indicative as naive chess players, including myself until my recent research, believe. When people have played several hundred rated games, that number starts to mean something.

For the record, I'm a Smyslov fan too.

Ubik42
DrCheckevertim wrote:

Cheating is actually relevant to the topic. If 10% of people cheated on chess.com (just as an example), that is a strong factor in rating pool deflation.

People cheating has zero effect on rating inflation/deflation.

redchessman

Its been a while since i've revisited this topic, but assuming we are comparing blitz rating to uscf standard:  The only correlation you can possibly make is that higher rated uscf players will in general have higher rated blitz ratings, you can't make a formula though because you are comparing long time controls to extremely short time controls.  Also not everyone is playing the same blitz time controls or long time controls.  For example, g/10 is blitz on chess.com and so is g/3.  While uscf standard ratings are based on anything greater than g/30.  However, when I play uscf tournaments, I limit myself to mostly g/90+30, 115d10 sd30, or 40/115 sd/1 d5.  This means my rating will most likely be different than if I played only g/45s.  Similarly in blitz, I am mostly playing 3 minute games and if I suddenly switched to 10 minute games which would still be blitz my rating could be much different.  When I had a good connection and played rated blitz I was 2300 with like a 1900 uscf.  Now i'm over 2100 uscf.  If you are comparing straight numbers, my blitz rating is still much higher than my uscf and we could say that would be classified as inflated. Finally, I think blitz rating depends on the type of player you are.  I know my openings fairly well so i can easily get a time advantage in 3 minute chess.  Then all i need to do is keep the position simple so there is no blundering and I can win by time or of course they can mess up and lose on position.  

Ok now some people are saying why are there NMs who are only like 2000 blitz. Some people say they have connection problems or whatever.  That can be a valid reason, but also I think you need to look at who those Nms are.  Many older NMs are very slow and you can flag them most of the time in blitz(3 minute).  If you look at younger NMs you'll notice they have higher blitz ratings.  

In general, without putting any formulas into the question, we can state the factors that make up higher rated blitz players assuming they already have a uscf: high rated uscf, strong opening knowledge, being young, and good internet connection.  

Now AdamRinkleff is making a point that chess.com ratings are deflated against uscf and he gives himself as proof.  He is 1850 uscf and 1470 blitz.  However, maybe his poor blitz rating can be attributed to factors such as being old (he has a phd apparently), weak opening knowledge, or poor internet connection.  Because honestly I don't understand how any competent 1850 uscf player can be 1400 blitz when my 1k uscf brother has higher blitz than that...  

Ubik42
zborg wrote:
bigpoison wrote:
 

Yes, simple techniques, but IF the point is to win, THEN how can there be engine users at all ratings levels?

Why would someone cheat occasionally to maintain a 1700 chess.com rating.  It doesn't make any sense.

If you are here to improve your playing strength, win games, and avoid detection, it makes perfect sense to use engines intermittently, get a 1.5 pawn advantage during some part of the game and boost your rating while still making the lion's share of your moves.

You typically end up with a rating roughly 200-300 points higher than you would have otherwise.  And you keep practicing and improving along the way.  Except you're using a crutch to win games, more often.

Only a "monkey" lets the engine make all the moves, and wins nearly all the time.  Yes, those ratings go sky high.  But that Monkey eventually gets caught, and they learn essentially nothing along the way.

Why does this sound unreasonable to you ??..

it is unreasonable because as soon as you win some games by cheating a little, and artificially raise your rating, you will be playing tougher competition. In order to continue having a high winning percentage, you will have to cheat more. And more. And more. And soon, you will have to cheat 100% of the time. Its inevitable.