Kasparov is slaughtering Nakamura and So.

I've literally just read this post. Not a single other post in this entire thread so correct me if I've misunderstood whats happening. Jen, are you saying that Magnus has more dominance over somebody 100 (for exmaple) points lower than him than a 1500 does over a 1400?

Assuming that your definition of dominance = % score vs them then you're completely 100% wrong. Of course he has the same dominance. He'll obviously have a far higher win to loss ratio because of the large number of draws, but the percentage score will need to be identical to maintain the ratings as they are in both cases.

Jengais seems unclear on the concept of math. 

If Carlsen regularly loses to lower rated players, they won't be lower rated than him anymore. 

There are Tournament Performance Ratings (TPRs) that track individual performances. Carlsen's rating is so astronomically high that even the top players will have career days just to match his *average* rating. 

Carlsen is expected to score ~2/3 (64% to be precise) against a player rated 100 points below him. That doesn't mean he has to win two and lose one, it means he can win one and draw two. In fact, if that 2750 player has White twice, his chances go up significantly. 

An established 1500 would be expected to score 4/6 against a field of established players rated exactly 100 points below him. However, there probably won't be any draws in those six games. 

Ratings really do work. They are actually measures of past performance, but the past is prelude. That is, ratings really do have a strong predictive power. That's why really smart chess players have been using ratings to determine relative playing strength for more than 40 years. 

Here's a link to an easy-to-use elo table:

http://www.pradu.us/old/Nov27_2008/Buzz/elotable.html

Jengais seems confused about a lot of things. Too much effort (for me) to talk it all out.

I've never played OTB. I understand maths though. Six wins, three losses and a draw may seem less dominant (and could well be intrepreted that way) than three wins and seven draws but they're both 65%. Your experience must be deceiving you.

DjonniDerevnja wrote:

heine-borel wrote:

A 250 gap at 1400 means that Magnus would win more than 80%, appx, of the points. 

It's still a 50 gap. 

(For reference, 200 points means about 75% of the score)

A 50 gap is circa nothing at 1400, and a huge difference at 2800. The ratingscore is about what happend in previous game. The next game is a completely different thing. Many 1400s looks at 1450s as players they shal beat.

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

Many 1400 players are actually performance rating wise between 1200 and 1900

While many 2700 are performance wise between 2650 and 2850

"Isn't it incredible to think that Carlsen would be expected to score 5 wins and 5 draws in a ten game match against a 2650?"

Not really, he has 200 elo easy over a 2650, and his 2850 kind of represents the peak of human ability, so he is less likely to miss anything, as opposed to a 2400 vs a 2200, against Carlsen it would be even more a certainty.

Everything you're saying is super obvious. I haven't got the slightest clue why you think I don't know it, why you think it's only obvious to people who play OTB or why it's even slightly relevant.

Taking that into account in a match (lets say 10 games) between a 1500 and a 1400, compared to a match between a 2700 and a 2600 or whatever else and you can say that the 1400 is more likely to win than the 2600. No shit. That was already true anyway because of the higher percentage of decisive games. This just makes it more true.

The average result will still be the higher rated player scoring ~64% vs the lower in both cases. The lower rated example will have more "upsets" and also more thrashings but the average outcome will be ~64%.

I am actually have a very good idea what I'm talking about. It's because of this thing I have called intelligence.

Why doesn't Garry come back to chess? Life without chess must be boring, and it doesn't seem that he'll ever become a successful politician. Maybe in the US...

What you write about that the higher rated will be higher scoring than the lower is less true OTB than online. 

The reason is that onlinerating is more up to date, and that the frequense of games usually is higher online.

When we play here on chess.com most of us gets the feeling that a 100+ player is stronger. OTB is different. OTB a much higher gap is needed to feel weaker. 

OTB rating is not  up to date. Myself I feel my OTB-rating is reflecting my strenght more than a year ago, while the onlinerating is pretty close to my level today.

Among the below 2000 there are a lot of players, many of them kids, that is climbing so fast that they often performs several houndred points above their rating.  A brilliant 13 years old kid  have a otb-rating that was very much lower two years ago, but those old games still drags his rating a bit down. On the internet people have much higher numbers of games, and therefore older games counts less.

What kind of an idiot thinks that it's fair to have doubt about the accuracy of the rating of the 1400 but not of the 1500? If you can't see that they virtually cancel out and aren't worth mentioning then... I don't know... I just don't know...

Lasker was a professional mathematician and world chess champion.

More recently John Nunn was a mathematical prodigy, the youngest professor at Oxford and a world top 10 chess player.

Elroch, you know that's not what he meant. 

It's true, lower rated players are less consistent than higher rated players. But if a player has an established 1500 rating and is playing a field of six 1400s, he or she will be expected to score 4/6. That's precisely the same score that a 2850 will be expected to score against a 2750. 

You have a point her, but still I am that kind of idiot. I believe that the more you approach the most common level for experienced players, which I think is around 1700 Fide, the more stable the ratings are in general. So 1500 is more stable than 1400.

I have written all the time that 1400 is very unstable because it is a step on the ladder where few players takes a long rest. 1500 has something of the same, but less. 1000,1100 and 1200 looks even more unstable, while 1800 is close to cemented.  I am talking about otb-fiderating.

Djonni, the average USCF rating is 1068. 

http://archive.uschess.org/ratings/ratedist.php

FIDE doesn't make finding out the average rating easy. I don't know if it's published anywhere. Of course, it will almost certainly be higher than USCF if only because it's a bit more difficult to get a FIDE rating. That means that only the most competitive players will have FIDE ratings. Once upon a time, the lowest FIDE rating possible was 2000. Now, the rating list is much more open. 

Wow !! The average USCF has really taken a hit with the explosion of scholastic chess !  Maybe we should only use adults to calculate average rating ?  

Reb, non-scholastic average ratings is only 1198. I'll have to go back and see what the historic ratings were, but I don't recall the average rating ever being 1500 for USCF.

I understand that. That is exactly the reason I used the word "virtually" and didn't just outright claim that they directly canceled out.

I could be mistaken but I thought back in the 70s the average uscf rating was 1500-1600 ? 

You are mistaken NM Reb....(sorry, I had to........)