July 07 Rating List

"every time the Petroff loses, an angel gets its wings."

That put me laughing, speaking of the Dortmund games. :-)

As MC noted, Adams appears twice (also with 2687 at #25). The reason seems to be that FIDE created a new ID for his MTel performance (2687) - Sofia is not included in his -3 since April.

After correction, Adams should fall behind Grischuk at #15.

Next Danailov press release: "Michael Adams #14 and Michael Adams #25 are the same person!"

Certainly looks like Jan Smeets got migged.

I cannot understand how FIDE could release a list with such a glaring error concerning Adams. They should have fix this before officially releasing it.

I also notice that Karpov is no longer on the list. I guess the tournament he just played in won't appear until the next list.

"I also notice that Karpov is no longer on the list. I guess the tournament he just played in won't appear until the next list."

One wonders if the fact that he was due to be dropped from the ratings list was an inducement for Karpov to reactivate himself. On the other hand, he's always had close ties to the Serbian chess community."

"Certainly looks like Jan Smeets got migged."

Indeed! Mig has a bit of a trope towards twisting proper names into verbs. Haha, look at the funny Dutch names, eh?

I'm sure that Kramnik would have liked to have won, but undoubtedly he was compensated generously for the putting those ELO points at risk.

Let's keep things in perspective: it's not as if Elite players get a bonus for each rating point, or each step up the ranking latter. Has Anand's chess career suddenly gotten a lot more lucrative now that is is the top rated player? Not really...it's pretty much the same as before.

Perhaps Adams' rating anomaly, stemming from the M-Tel event in Sofia, has also impacted Topalov's rating, possibly adversely?

"Jakovenko hits the top ten, replacing his compatriot Svidler and validating Alex Yermolinsky's eye for talent. ["The tournament introduced a lot of strong young players. Jako is already a Top Ten material, no doubt about that. Nepo is a major talent, on par with Carlsen, Karjakin and Nakamura." Posted by: Yermo at December 18, 2006 12:30] Unfortunately his poor showing at Aerosvit probably means he won't be there for long."

Jakovenko has suffered lopsided losses in a manner that no Top 10 player should allow. Clearly, he is a 2700 ELO player (although that benchmark has become less meaningful in the past 2-3 years. Now there are 8 players who are 2750 or higher, with Radjabov at 2746. So, pretty soon, 2750 will be the new 2700.

"Wang Yue of China moved up to #22 with 25 games played. His 2696 is 40 points over his last rating of just three months ago, a great leap forward indeed."

Somebody ought to scour his tournament record, and see if he's played a couple of big tournaments in Myanmar recently. It almost makes one forget the name of the previous "Best player from China".... Bu Who?? If Wang wants to be considered to be a Top 20 player, with a solid rating of (nearly) 2700, then he need to trek over to Europe and accept some invitations to strong events, and put those points at risk.

Of course, when one considers the overwhelming success of the Chinese Women in the Women's World Team Championship event a few weeks ago, it may not even be the top achivement that the "Chess School of China" has produced

One important thing to note that over the last 10-15 years, many players have found that reaching #4 is the farthest they could go. With Kaspy-Anand-Kramnik (and lately Topa) monopolizing the top 3 places, we have seen, so many players, Adams, Svidler, Moro, Mamedyarov, Aronian to climb to #4 or #3 but not beyond that!

Aronian seems to be the best bet to climb to #2 or higher. After great Karpov decided to become an amateur, when was the last time there was a player [not named Kaspy, Anand, Kramnik, Topa] to reach #2 or higher. Well, the new generation of Aronian, Shak, Radja and Carlsen have to do a lot of hardwork!

Immediately after the post above, I realized that I have missed Ivanchuk's name who was a real heir to Kaspy;s throne. It's just that in spite of being as good as big 4 [Anand, Kramnik, Topa, Leko] his erratic results affected his rating.

Amit,
Jakovenko and Karjakin look also very promising, IMHO. And don't underestimate Cheparinov, the fourth ego of Mr. Danailov (after himself, Pono, and Topa)

Mig, I remember you saying that Maxim Sorokin is your old friend. My condolences: he died today in a car crash.

"Jakovenko replaces Svidler" - wait, aren't Jakovenko, Shirov, and Svidler tied for 10-12? Aren't they just listed in English alphabetical order?

They are listed by number of games played

Sad to see Korchnoi out of the top hundred, but all things must pass. Wouldn't surprise me if he's back on the next one. The man is a geriatric wonder.

Chucky has a brand new 2762 rating and that does not include the bundle of rating points he is set to acquire for winning first outright at Foros with his +4 =7 -0 performance.

So Chucky's new effective rating is around, say, 2775. Thus he has leaped over Topalov for World #3. Doubt that he has passed Kramnik since the WC is having an excellent +3 =3 -0 Corus as we speak.

Hey Mig...I think you need to make another post on FIDE Folies III. How about this "shake that hand lose rule"? Your thoughts on it?

- TCG

How about Wang Yue? "Watch yer step, varmint. I'm gonna Wang Yue" or "Oh my precious pudding pie, how I wish I could Wang Yue"

DOug, you wrote about Wang Yue:

"Somebody ought to scour his tournament record, and see if he's played a couple of big tournaments in Myanmar recently. It almost makes one forget the name of the previous "Best player from China".... Bu Who?? If Wang wants to be considered to be a Top 20 player, with a solid rating of (nearly) 2700, then he need to trek over to Europe and accept some invitations to strong events, and put those points at risk."

http://fide.com/ratings/tourarc.phtml?codt=24&field1=8601429 shows the tournaments of Wang Yue. He won 13 points in Calvi (France), 15 in Cappelle la Grande (France) and 12 on the Philippines. No Myanmar to be seen. In Cappelle la Grande he played 8 players with an elo of 2526 or more. It seems he already made the trek over to Europe and put his elopoints on risk.

That we haven't seen him in a strong European round robin is logical; those are reserved for the incrowd. Just pick the average 2650 player, without a big name like Karpov, and check how many strong invitationals he got to play the last five years outside his own country.

Presumably this Bu Who? of whom Doug speaks would be the Bu Xianghi who led China into - was it silver? bronze? - medal position at the Olympiad, unbeaten IIRC on top board and drawing with Kramnik.

amit,ratings are important,but the big guns don't care that much about them like us,the mortals.had morozevich protect his gain,he wold be no 2 in the world now at 2773,but he played bosnia tourney(probably going for no 1!!)and lost 14 points scoring 50%maximalist like allways!debates like:nobody was able to break the trio kram-anand-topa are pointless(still i apreciate the consitency of kramnik and anand)

I can't understand how the Michael Adams error came about and certainly it should have been caught. I wrote directly to the address on the FIDE page too pointing it out. If I was in the FIDE office I'd have already corrected it, we all make mistakes, but the point is to correct them as soon as possible.

At the moment the top list is worthless as you simply don't know the impact an unrated player, rather than a 2700 one, competing at MTel will have on other ratings.

Korchnoi's dropped out of the top 100 several times before, its not necessarily the passing of an era as he's worked his way back.

As to Karpov he's been inactive for a few lists now but will be back in the top 40 after his performance in Valjevo.

None of these guys can beat Rybka so how good are they?

None of the current marathon runners can beat a Porsche so how good are they?

The tournament report of the M-Tel tournament:

http://fide.com/ratings/trarc.phtml?event16=7986&codt=24

Several things can be noticed:

- The average elo of opponents (the column with Rc) is wrong, therefore the calculations are wrong as well.

- The first name of Topalov is usually spelled with only one s, not as "Vesselin".

When clicking on the name of a player, we get to see the results of that player in this rating period. It is now obligatory to give the results for every single game, because elo gets calculated for every game. However, for the M-Tel tournament only the average elos can be seen. An example: http://fide.com/ratings/tourarc.phtml?field1=13401319&codt=24

A possible explanation is that the tournament was not reported properly to FIDE, and that FIDE created the rating report itself, making some errors.

Not a good analogy, Rybka's "only" a 130 points better; what's Porshe's top speed?

hetman,to be polite,go ..and chat with the computer programs!

Yeah, let's decide on a formula to convert elo to mph.

Sorry to hear the news about Maxim Sorokin.

On a Russian forum they say he died in an Elista hospital, as the car crash happened while he was on his way to Volgograd from Elista. He was Rublevsky's second at the Candidates.

Hi,

following our discussions after lots of non-playing draws at the Euro Ch. I decided to do some bean-counting as soon as the new list arrives. And here is the result:

From the 98 GM-level players currently in the Top 100 List the one least likely to make a short, fightless draw is >>> David Navara <<<.

If tournament organizers look for some value for their money, here they are:

Based on chessgames.com:

Least likely to draw in 30 moves or less (either color):

David Navara CZE 6.54%
Hikaru Nakamura USA 7.77%
Baadur Jobava GEO 7.98%
Darmen Sadvakasov KAZ 7.99%
Alexander Morozevich RUS 8.87%
Krishnan Sasikiran IND 9.25%
Wang Yue CHN 9.40%
Arkadi Naiditsch GER 9.50%
Vladimir Malakhov RUS 9.52%
Gata Kamsky USA 9.73%

Least likely to draw with white in 15 moves or less

David Navara CZE 0.00%
Wang Yue CHN 0.00%
Arkadi Naiditsch GER 0.50%
Judit Polgar HUN 0.58%
Pavel Eljanov UKR 0.61%
Daniel Stellwagen NED 0.70%
Darmen Sadvakasov KAZ 0.77%
Pavel Smirnov RUS 0.85%
Sergey Karjakin UKR 0.87%
Alexey Shirov ESP 0.89%

And on the down side, the candidates for the Special Xtreme Drawnik Award

Most likely to draw with white in 15 moves or less:

Borki Predojevic BIH 12.15%
Konstantin Landa RUS 11.72%
Konstantin Sakaev RUS 11.21%

Most likely to draw in 30 moves or less (either color):

Alexander Khalifman RUS 38.09%
Vadim Zvjaginsev RUS 36.11%
Konstantin Sakaev RUS 31.64%

"Korchnoi's dropped out of the top 100 several times before, its not necessarily the passing of an era as he's worked his way back."

Kortchnoi has been close to the cut-off (of being in the Top 100, or just missing it) for the past few years. It should be noted that the rating of the #100 player (whoever occupies that slot) has been rising by a few points every rating list. Kortchnoi's rating varies up and down, but every list the rating of the 100th player edges closer to the upper value of Kortchnoi's rating range.
I do hope that he makes it up into the top 100 again.

-----------
"Presumably this Bu Who? of whom Doug speaks would be the Bu Xianghi who led China into - was it silver? bronze? - medal position at the Olympiad, unbeaten IIRC on top board and drawing with Kramnik."

Silver Medal, I think it was. Certainly, both the Chinese Men and Chinese Women have formidable records in team competitions. There is no doubt that Bu Xianghi is a very strong chessplayer. However, for once having held the record for earning the GM Title at the youngest age, his vector of success has flattened. Having been born in 1985, he is now too old to be included in the Top (20) list of Junior players. He now must solely be measured against the best players in the world. Now that Wang Yue has supplanted him as the
Chinese player with the best ranking/highest rating, Bu risks falling into obscurity. He could well miss out on the one slot that many European organizers offer/reserve for a Chinese player...in Invitational events. Hence, "Bu Who?".

-----------

"That we haven't seen him in a strong European round robin is logical; those are reserved for the incrowd. Just pick the average 2650 player, without a big name like Karpov, and check how many strong invitationals he got to play the last five years outside his own country."

The Chinese contingent went to both Cappelle la Grande and Calvi, and generally excelled. I believe that it is precisely because these events offered good prospect for success that the Chinese
federation organized the chess excursion.
In any case, I hope that Wang Yue gets a chance to
test his mettle in a strong Invitational (Round Robin) chess events. Chess phenoms have tended to have a career trajectory of shredding the opposition in Open Swiss system events, gaining a boat load of ELO points, then garnering invitations to strong Round Robins on the basis of
their attractive, almost tumescent, ratings. Once they play in the Round Robins, they tend to plateau. Indeed, their ratings can stagnate, and sometimes will drop about 50 points as they struggle to adjust to the more difficult and challenging tournament environment.

Maybe Wang Yue IS the next Mamedyarov (who seems like the real deal, having "ratified" his ELO on the basis of several results)--but I won't
be convinced until he runs the gauntlet.

I noticed for ratings calculation the details on the level of single games for Morelia\Linares 2007 are also not listed. Is it a valid method to calculate based on average opponent rating ? For this can only be an approximation (as has been said those given for M-Tel 07 are wrong). I calculated ratings impact of M-Tel 07 based on Apr 07 ratings list using both methods (actual / average opponent rating):

Topalov -2.98 / -3.02
Mamedyarov -5.44 / -5.48
Adams -6.52 / -6.52
Sasikiran 6 / 6.04
Kamsky 3.45 / 3.48
Nisipeanu 5.49 / 5.53

One caveat: unknown to me is which numbers they round in which way during calculation.

FIDE has switched to the new method of calculating rating changes game by game rather than using the average opposition rating over a tournament last year, but for some reason they already applied the old method for Linares. I expected them to correct this later, but they haven't. Same for MTel now, although there seem to be other things wrong too with the MTel calculations. Btw, it seems like Karjakin's Elo gain from the Ukraine Games in May is missing. Wasn't this a rated event?

Another thing I've been wondering about is, what is the official deadline for submitting rating reports now? One would have expected FIDE to learn from the Linares fiasco in the April list, but they've done the same thing as before: first announced a May 31 deadline and then ignored it by including June events like the candidate matches.

Bartleby, I've been calculating a ranking of the players with the fewest short draws for a while. See my chess statistics website: http://members.aon.at/sfischl/stat.html
The results are quite similar to your numbers.

Smeet as a unit of measure reminds me of a smoot, the unit of distance employed by MIT students to render the length of the Harvard Bridge as "364.4 smoots and one ear" - see http://en.wikipedia.org/wiki/Smoot

Wow, I never dreamed I'd ever hear the "364.4 smoots and one ear" story again -- let alone on Wikipedia!!!!

Vividly I remember hearing it told to me during my freshman year of college (1972), the first time I saw the odd markings on that Harvard Bridge. The details I heard were exactly as in the Wikipedia entry, except I recall being told that the unfortunate Mr. Smoot passed out drunk at a party and his mates then marked off the bridge using his unconscious body; rather than his doing it voluntarily as part of a pledge ritual. (Not a major difference.)

It's great, I guess, that they carry on the tradition and re-paint the markers periodically so they don't disappear. I'd assumed those markings, and the story too, had withered away long ago.

littlefish/bartleby

The problem with the otherwise great littlefish list, is that it doesn't take the quality of the opponents (relative to the players themselves) into account in a good/sufficient way. That Yue Wang won't offer draws to players 200 points below him, should come as no surprise, for instance.

Oscar/DOug

Regarding Wang "risking his points", I don't think neither he nor the other 2 rising Chinese players (Bu and Ni) have done that against reasonable opposition during this quarter - consider the following excerpt from one of my postings on cg.com (ratings for januar, april and july in the table below):

Wang, Yue........ g CHN 2644 2656 2696 (+40)
Bu, Xiangzhi ..... g CHN 2644 2656 2685 (+29)
Ni, Hua............. g CHN 2632 2654 2681 (+27)

Bu and Wang have mainly been deadly efficient in beating 2500 GMs and lower rated players (both played some 2300-2400-players in low profile tournaments).

The score for Wang and Bu
(against below 2600):
Wang: +14 =3 -0, Bu: +16 =5 -0) -> +30 =8 -0,
(against above 2600):
Wang: +1 =7 -0, Bu: +1 =4 -1) -> +2 =11 -1.

Yue Wang "only" managed to beat his countryman Hao Wang (2638) out of his 8 opponents above 2600, while Bu beat Zoltan Gyimesi (2616) out of his 6. 89% score against their below 2600 opponents is as if all their opponents were rated 2300 (based on the 2650-ratings of Bu and Wang), so +30 =8 -0 is pretty extreme... :)

Hua Ni has a similar profile and played the same 3 tournaments as Wang. In the 24 rated games he's got against below 2600, he scored +14 =7 -0 and against above 2600, +0 =3 -0. All those 3 were chinese btw, and he lost 0,3 points from those games, while he gained 26,9 against his below 2600 opponents. For all three of them, these stats are:

Against below 2600 above
Wang .. +36,9 ____ +2,8
Bu ...... +30,4 ____ -1,7
Ni ....... +26,9 ____ -0,3

How strong these 3 chinese players really are, probably needs to be proven against stronger opposition, where 2650+ opponents should be more than enough so far - it would mean a radically tougher obstacle for them than who they played now. While 17 games for 3 players isn't that much, a total of +0,8 in those 17 games doesn't indicate a strength much above 2650. [Bu: +2616, =2667, -2750, =2645, =2627, =2622, Wang: =2604, +2619, =2632, =2632, =2658, =2602, =2621, =2654, Ni: =2644, =2621, =2656]

FIDE have now issued a corrected list with Topalov in 2nd place (with the same rating as Kramnik, but having played more games during the rating period).

There is only one Michael Adams in the Top 20 now, and he is 15th with a rating of 2724.

Anyone else notice the rise of Vladimir Afromeev? For a mere FM, he seems to be doing well.

> Anyone else notice the rise of Vladimir Afromeev? For a mere FM, he seems to be doing well.

Yes he SEEMS to be doing well. I've analysed the performance that he put in to increase his rating in the latest TWIC.

http://www.chesscenter.com/twic/twic660.html

People interested in the "phenomenon" Afromeev, might want to have a look here, too:

http://www.chessgames.com/perl/chessplayer?pid=66272&kpage=4

When the previous ratings list came out, I got an idea of trying to measure how well the various players rated 2700+. I tabulated total number of wins, losses, assorted percentages and mini-match scores on the basis of the last 15 months of performance. The result was a fun little trivia hunt where I gave a description of each player's performance and people tried to guess what player I was talking about (for example, most guys guessed that Kramnik would have the best record--a lot fewer figured out that Kamsky would have the lowest draw percentage).

This time, no game, sorry. But some very interesting results.

So again, what I did, was take every GM whose rating is currently at 2700 or higher (that means only one Mickey Adams) and determine his head to head results against other 2700s.

Here is the best of results.

If you would like the complete list, please e-mail me:

my last name (at) wife of jupiter, that is, juno (period) com

(for all of these rankings, I omitted the three GMs who played less than 10 games against other 2700s, ie Polgar, Akopian & Eljanov)

Number of wins:
1. Topalov (12)
2. Aronian (11)
3. Gelfand (10)
Total number of games played:
1. Aronian (57)
2. Svidler (55)
3. Topalov (51)
Percentage of wins (#wins/#games)
1. Kramnik (28.1)
2. Gelfand (24.4)
3. Topalov (23.5)
...and lowest
19. Radjabov (10.7)
20. Jakovenko (8.3)
21. Shirov (7.0)
Percentage of losses (#losses/#games)
1. Kramnik (6.3)
2. Radjabov (7.1)
3. Mamedyarov (9.4)
...worst percentage of losses
19. Topalov (25.5)
20. Morozevich (31.4)
21. Jakovenko (33.3)
Decisive percentage (#wins + #losses/# games)
1. Morozevich (54.3)
2. Topalov (49.0)
3. Kamsky (45.5)
...and lowest percentage of decisive games
19. Leko (27.3)
20. Adams (22.2)
21. Radjabov (17.9)
Scoring percentage ((#wins+.5#draws)/#games)
This is an average number of points you score per game (1 for a win, half for a draw, 0 for a loss). Another way to think of it is if you played 100 games with the same success you normally have against 2700, how many points would you score? Answer is rounded to a nearest half-point of percentage.
This to me is one of the more informative results so full list is supplied here.

Kramnik 61
Gelfand 56 (weren't expecting this one, were you?)
Ivanchuk 55.5
Mamedyarov 54.5
Aronian 53.5
Anand: 52.5
Radjabov 52
Leko 51.5
Adams 50
Topalov, Svidler 49
Ponomariov 48.5
Carlsen 48
Morozevich, Kamsky 45.5
Grischuk 44.5
Shirov 42
Jakovenko 37.5

Akopian, Polgar, Eljanov all played less than 10 games against other 2700s.

Mini-match records (how many 2700+ each 2700+ has winning/losing/even record against over the past 15 months):

Anand +5-4=6
Topalov +4-5=6
Kramnik +7-0=8
Ivanchuk +5-2=10
Morozevich +3-4=7
Mamedyarov +4-1=12
Leko +3-3=9
Aronian +7-4=5
Radjabov +1-2=11
Jakovenko +1-3=4
Shirov +1-7=10
Svidler +5-6=9
Gelfand +5-4=7
Grischuk +2-3=7
Adams +1-1=9
Kamsky +3-2=4
Carlsen +3-7=7
Akopian =2
Polgar +1-1=1
Ponomariov +4-5=8
Eljanov -2=4

I then got another idea. What if you were to come up with a ranking based on who you have defeated and who has beaten you. Here is the system I came up with:

For each GM you have a winning record against you get an X number of points based on his FIDE ranking. You get 21 points for having a winning record against Anand, 20 against Topalov, etc., all the way down to 1 point for having a winning record against Eljanov. On the other hand, for each GM you have a losing record against you lose an X number of points, based on the same principle, but in reverse, the higher ranked the player, the less points you lose. So you lose one point for having a negative record against Anand, 2 against Topalov...etc. I named these points chess-dinars.

I feel that this is a fascinating measure of relative strength and helps to answer a question--how good is the guy who beats Kramnik, gets squashed by Leko and can only draw Ponomariov? Well, one version of the answer is here. It is one way to evaluate guys who are capable of trumping some of the top players but have intermediate success overall as well as guys who blow out the lower-ranked competition but stumble against top GMs. Also, unlike the rankings above, it doesn't allow you to acquire additional points for more victories over the same GM. So if you are Kaspy, you aren't ranked higher because you keep beating Shirov. And if you show potential with a victory over a top-ranked player, it will compensate for you not having been equally successful against everybody in the world. And a very strong player may stumble against one guy but compensate for it by winnning against many others.

Here is the number of chess-dinars for each 2700+

Kramnik 95
Aronian 52
Ivanchuk, Mamedyarov 33
Anand 20
Kamsky 18
Polgar 14
Gelfand 13
Leko 12
Ponomariov 7
Akopian 0
Svidler -2
Topalov -4
Adams -8
Radjabov -10
Eljanov -14
Grischuk -17
Morozevich -18
Carlsen -22
Jakovenko -33

If you read through all this and found it at all interesting, I hope you will e-mail me asking for the full crosstable, which includes individual records. Especially if you have access to your own database and could verify my results :)

Enjoy and I will try to do another one in three months.

When the previous ratings list came out, I got an idea of trying to measure how well the various players rated 2700+. I tabulated total number of wins, losses, assorted percentages and mini-match scores on the basis of the last 15 months of performance. The result was a fun little trivia hunt where I gave a description of each player's performance and people tried to guess what player I was talking about (for example, most guys guessed that Kramnik would have the best record--a lot fewer figured out that Kamsky would have the lowest draw percentage).

This time, no game, sorry. But some very interesting results.

So again, what I did, was take every GM whose rating is currently at 2700 or higher (that means only one Mickey Adams) and determine his head to head results against other 2700s.

Here is the best of results.

If you would like the complete list, please e-mail me:

my last name (at) wife of jupiter, that is, juno (period) com

(for all of these rankings, I omitted the three GMs who played less than 10 games against other 2700s, ie Polgar, Akopian & Eljanov)

Number of wins:
1. Topalov (12)
2. Aronian (11)
3. Gelfand (10)
Total number of games played:
1. Aronian (57)
2. Svidler (55)
3. Topalov (51)
Percentage of wins (#wins/#games)
1. Kramnik (28.1)
2. Gelfand (24.4)
3. Topalov (23.5)
...and lowest
19. Radjabov (10.7)
20. Jakovenko (8.3)
21. Shirov (7.0)
Percentage of losses (#losses/#games)
1. Kramnik (6.3)
2. Radjabov (7.1)
3. Mamedyarov (9.4)
...worst percentage of losses
19. Topalov (25.5)
20. Morozevich (31.4)
21. Jakovenko (33.3)
Decisive percentage (#wins + #losses/# games)
1. Morozevich (54.3)
2. Topalov (49.0)
3. Kamsky (45.5)
...and lowest percentage of decisive games
19. Leko (27.3)
20. Adams (22.2)
21. Radjabov (17.9)
Scoring percentage ((#wins+.5#draws)/#games)
This is an average number of points you score per game (1 for a win, half for a draw, 0 for a loss). Another way to think of it is if you played 100 games with the same success you normally have against 2700, how many points would you score? Answer is rounded to a nearest half-point of percentage.
This to me is one of the more informative results so full list is supplied here.

Kramnik 61
Gelfand 56 (weren't expecting this one, were you?)
Ivanchuk 55.5
Mamedyarov 54.5
Aronian 53.5
Anand: 52.5
Radjabov 52
Leko 51.5
Adams 50
Topalov, Svidler 49
Ponomariov 48.5
Carlsen 48
Morozevich, Kamsky 45.5
Grischuk 44.5
Shirov 42
Jakovenko 37.5

Akopian, Polgar, Eljanov all played less than 10 games against other 2700s.

Mini-match records (how many 2700+ each 2700+ has winning/losing/even record against over the past 15 months):

Anand +5-4=6
Topalov +4-5=6
Kramnik +7-0=8
Ivanchuk +5-2=10
Morozevich +3-4=7
Mamedyarov +4-1=12
Leko +3-3=9
Aronian +7-4=5
Radjabov +1-2=11
Jakovenko +1-3=4
Shirov +1-7=10
Svidler +5-6=9
Gelfand +5-4=7
Grischuk +2-3=7
Adams +1-1=9
Kamsky +3-2=4
Carlsen +3-7=7
Akopian =2
Polgar +1-1=1
Ponomariov +4-5=8
Eljanov -2=4

I then got another idea. What if you were to come up with a ranking based on who you have defeated and who has beaten you. Here is the system I came up with:

For each GM you have a winning record against you get an X number of points based on his FIDE ranking. You get 21 points for having a winning record against Anand, 20 against Topalov, etc., all the way down to 1 point for having a winning record against Eljanov. On the other hand, for each GM you have a losing record against you lose an X number of points, based on the same principle, but in reverse, the higher ranked the player, the less points you lose. So you lose one point for having a negative record against Anand, 2 against Topalov...etc. I named these points chess-dinars.

I feel that this is a fascinating measure of relative strength and helps to answer a question--how good is the guy who beats Kramnik, gets squashed by Leko and can only draw Ponomariov? Well, one version of the answer is here. It is one way to evaluate guys who are capable of trumping some of the top players but have intermediate success overall as well as guys who blow out the lower-ranked competition but stumble against top GMs. Also, unlike the rankings above, it doesn't allow you to acquire additional points for more victories over the same GM. So if you are Kaspy, you aren't ranked higher because you keep beating Shirov. And if you show potential with a victory over a top-ranked player, it will compensate for you not having been equally successful against everybody in the world. And a very strong player may stumble against one guy but compensate for it by winnning against many others.

Here is the number of chess-dinars for each 2700+

Kramnik 95
Aronian 52
Ivanchuk, Mamedyarov 33
Anand 20
Kamsky 18
Polgar 14
Gelfand 13
Leko 12
Ponomariov 7
Akopian 0
Svidler -2
Topalov -4
Adams -8
Radjabov -10
Eljanov -14
Grischuk -17
Morozevich -18
Carlsen -22
Jakovenko -33

If you read through all this and found it at all interesting, I hope you will e-mail me asking for the full crosstable, which includes individual records. Especially if you have access to your own database and could verify my results :)

Enjoy and I will try to do another one in three months.

Gladiator,

It is funny how you belittle the accomplishments of the Chinese players, members of a proven, world chess power... and you use a questionable statistical method to do it. You're not the only one. I've seen it done on the DD times before.

The Chinese players are strong... no question about it. You overlook their accomplishments in Olympiad (against top players), team events and top tournaments, yet you still question whether these 2650+ players are strong? How closely do you follow chess? Why is it that people always question the strength of players outside of Europe (namely China and India), but these questions are never raised with European players?

You are not to accept full blame for your bias. The general chess media coverage is not universal and not much is known about the regions outside of Europe and North America.

Gladiator made some reasonable points about 3 specific Chinese players. There are players who achieve high Elo by mostly scoring very well against lower players. Oleg Korneev is an interesting example, he plays very large numbers of games in open Swisses and is ruthless against 2400-ish players. Last year he was as high as 2671.

It is possible that the 3 Chinese players mentioned are absolutely thoroughly prepared in technical and sporting aspects of chess, such that they maximize their chances against technically inferior players. It is interesting and potentially useful to study how different players/ federations achieve whatever success they get.

gg,

I disagree. His implication is that these players are not as strong as their rating indicates and that they have reached the 2650+ plateau by beating a steady diet of weaker players. All his study showed was how efficient Chinese are at beating weaker players in a fixed number of games (17!)... not reasonable. It is a bias I see often in the chess press.

Do we realize that Chinese men won the silver medal in the 2006 Olympiad... and "gladiator: is questioning the strength of Chinese players?? Here's how they did it. Notice the strength of opposition.

http://schach.wienerzeitung.at/tnr3410.aspx?art=23&snr=12&lan=1&flag=30

Gladiator didn't "belittle" the Chinese players (although he did post a link critical of Afromeev, a European who deserves all the criticism he can get)- he merely pointed out that they haven't done all that much against 2600+ players in the last quarter.

In some ways, the link you posted supports both views - Bu Xiangzhi (at one time the youngest GM in the world) and Wang Yue were merciless on weaker opponents, but got a lot of draws against stronger players. Both only beat one 2650+ player - Kamsky for Bu, Van Wely for Yang. On the other hand, they did not lose any games to them.

2650+ players rarely give draws to players they think they can beat, especially with the White pieces. No doubt a few of these draws were arranged by captains, or the result of the stronger player being off-form or not getting a position they liked, but on the whole we can safely conclude that their opponents really thought they were that good. And they should be better placed to judge than any of us!

In any event, a superb Olympiad performance by the Chinese men's team.

James Raynard,

You are committing the same error that gg is committing. Somehow you trying to say the only reason the Chinese drew some games is because they were arranged or the other player had an "off day."

You didn't look at the chart closely (or didn't know how to dicipher it), but the point is these are strong players and there is NO doubt about their strength whatsoever. Wang Yue beat three 2650+ players and drew with Bareev (a former 2700) and Harikrishna (TPR of 2837). At the time, he was only 2598. Bu also beat Jobova and had a TPR of 2790.

No matter how you slice gg's stats, the point is beating 2600-2700s is not unusual for Chinese players... they can compete with the top players on equal terms.

Question: How do you get to 2650+ beating primarly 2400-2500s?

Answer: You can't.

Let me see if I can understand the incredibly complex Olympiad chart.

The Chinese team only won one match against a team rated higher than them (Netherlands, in the last round, when teams that don't already have prize spots usually go for broke).

The Zhangs both achieved an impressive performance of 0.5 against players with a ranking equal to them or higher.

Pengxiang Zhang faced an opposition which was on average rated 65 points below him and still got an Rp which was over 20 points below his rating.

Wang Yue's wins (with exception of one, again, against Van Wely, 2655, in the last round) all came against players ranked at least 38 below him (and who on average have a 2492 rating). The statement "Wang Yue beat three 2650+ players" is either an outright lie or a result of somebody's inability to dEcipher the chart.

Ni Hua only once faced a player whose rating was not less than 20 points below him once. Nonetheless he managed to perform ten points below his rating.

While perhaps beating 2700s is not that unusual for Chinese players, the highest rated player Xiangzhi Bu beat on his way to 2790 Rp was a 2671.

Only two players on the team achieved a rating performance higher than their rating.

All but Bu faced opposition which on average was rated lower than them.

If it's impossible to get to 2650+ by beating primarily 2400-2500, it is at least apparently possible to get to 2837 by drawing 2600s and beating 2400-2500s.

The third paragraph should read "The Zhangs both achieved an impressive performance of 0.5/4 against players with a ranking equal to them or higher."

Yuriy,

Sino-pessimism still reigns here.

OK... but what's your point? Are the Chinese strong or not? That's really the whole point. Maybe this is a bit simpler.

http://www.fide.com/ratings/topfed.phtml

Either the chart is lying, or China really has incredibly strong players. This case is closed.

"Sino-pessimism still reigns here...but what's your point? Are the Chinese strong or not?"

No, Daaim, this isn't. I am not judging Chinese people or even Chinese chess players as a whole. My point had to do with the Chinese Olympic performance chart, which you seemed to feel supports the claim that these are incredibly strong players when it comes to face to face results with players their own rank. Looking at it, that is definitely not the case. For two of them (the Zhangs) the performance is very far from good. Yue achieves almost all of his wins in the tournament against considerably lower-ranked competition. Ni never faces competition his own rank and still underperforms. Zhao beats 2 2200s and loses the only other game he has. The only one who can be said to perform well against players his own rank is Bu, who faces guys ranked above 2600 9 times and wins twice. Not bad, but not exactly San Luis material. Much like gladiator claims above, Wang and Bu's Rp is misleading since most of it comes from victories against lower-ranked competition. Again, it is not only possible to get to 2650 by beating 2400-2500s, but the chart suggests that it's possible to even get to 2837 by doing the same.

Considering the very start of this conversation was that the FIDE ratings can be misleading as far as actual player strength, it's very strange that you would use FIDE's country ratings to support the claim that Chinese top players are as strong as their rating suggests. It's kind of like if I said you had a fake passport and you tried to prove to me your identity by showing me the fake passport again.

"This case is closed."

Well, at any rate it's nice to know you at least convinced yourself.

Daaim,

Speaking for gg, I didn't give any stats to be sliced. "No matter how you slice gg's stats".

"Question: How do you get to 2650+ beating primarly 2400-2500s? Answer [Daaim version]: You can't."

My answer: By scoring 76% against 2450 opposition. Like Mr. Korneev.

If players X and Y have the same elo, say 2650, and X scores higher % against 2650.

Daaim, it is curious that immediately after several posts dissing Afromeev (and fully justified) you say "Why is it that people always question the strength of players outside of Europe (namely China and India), but these questions are never raised with European players?". Despite his name, we may surmise that Afromeev is a European player.

These blogs often comment on questionable Elo results of European players. How about Mr. Crisan of Romania? All the non-existent norm tournaments in Ukraine. You just don't notice these?

Gladiator's original claim was that "Bu and Wang have mainly been deadly efficient in beating 2500 GMs and lower rated players". He did not say that their elo's are undeserved.

It would be interesting if players such as Nakamura adopted Chinese-style sporting discipline. Maybe he would lose less games from weird openings, maybe he wouldn't enjoy chess anymore.

"Somehow you trying to say the only reason the Chinese drew some games is because they were arranged or the other player had an "off day.""

No I am not. What I actually said was:

"No doubt a few of these draws were arranged by captains, or the result of the stronger player being off-form or not getting a position they liked, but on the whole we can safely conclude that their opponents really thought they were that good."

Please pay particular attention to the words "a few" and the clause beginning with "but on the whole".

In other words, I was saying that most of their draws were neither pre-arranged nor the result of opponents having a bad day. i.e. almost the exact opposite of what you claimed I said.

"You didn't look at the chart closely (or didn't know how to dicipher it)"

One of us is clearly having seriously problems with this table (it's not a chart). I assumed, obviously completely erroneously, that the column head "Rtg" gave the opponent's rating, the column headed "Res." gave the player's colour (w=weiss, s=schwarz) and result for that round (1=win, half=draw, 0=loss). Perhaps you could explain how it should be interpreted?

Continuing regardless, we see that Wang Yue's victories were against Hakki(2459), McNab(2437), Flores(2494), Dimakiling(2452), Gagunashvili(2560), Laznicka(2551) and Van Wely(2655).

In summary, he beat one 2650+ player (Van Wely), not three, and did not even play against Jobava, let alone beat him.

"Question: How do you get to 2650+ beating primarly 2400-2500s?

Answer: You can't."

Correct answer: You can.

Afromeev got to 2642 by beating primarily 2100-2200s (Mark Crowther put one of his cross-tables on TWIC, as he mentioned earlier in the discussion). Dennis Bloodgood got to 2700+ by beating his fellow inmates.

And no, I'm not comparing Chinese chessplayers to dubious Russian businessmen or convicted murderers, just illustrating a peculiarity of the ELO system.

James Raynard,

I understand your points. You're right about the column in the Olympiad table, my mistake. That still doesn't make your claim that Chinese players are not tested as strong players. Considering that there are so few 2650+ players in any given tournament, I'm not sure what you want the Chinese to do. Can they invite themselves to Corus, Dortmund, Linares, the Bundesliga? They are not getting invitations to top tournaments. Those are reserved for the same permutation of 10-15 players.

You can get to 2650 by beating 2100s? Oh sure. Claude Bloodgood... oh. Crowther put ONE of Afromeev's tables up? Oh. I guess the one table proves it. I probably could get to 2700 if I arranged enough illegal tournaments. Is that the issue here?

As I said... refer to the FIDE country chart. Certainly one may argue that China is not as strong as Armenia, Georgia and a host of other chess powers, but there is something to be said about the rising tide of Asia.

Strangely enough, I had a debate with a person on the DD who said that China was not a chess power! He didn't know that the Chinese women had won (at the time) four Olympiads in a row (now five) and had three recent World Champions. He said they hadn't proven themselves!! This was before the Chinese men won the silver in the last Olympiad. Both the men and women are only getting stronger with a torrent of young talent in the pipeline. Wang Hao wasn't even on the silver medal team!

gg,

In general, European players are assumed to be legitimate until they are disproven. Non-European players are assumed to be illegitimate until they are proven. That's the difference. I've seen this chess arrogance in real life. Even Anand got that treatment in the early days. People did not believe that such talent could come from India and said he needed to be tested.

This notion will continue until China and India begin to dominate the top 50. Sheers numbers, inertia, focus and sponsorship will make this a reality. With China already #3 in the world rankings and India developing at a blinding pace, that time will come... and soon.

The issue, as I understood it, was that players can have ratings that do not reflect their real strength, and that the three Chinese players MAY be examples of this.

Afromeev and Bloodgood are extreme cases of blatant cheating, who should have been stopped by the authorities before things got completely out of hand. (No-one to my knowledge is claiming that the Chinese players in question have done anything improper).

As you only provided one crosstable to support your argument, I'm not sure why you're so scathing about Mark Crowther.

I wasn't claiming that these Chinese players hadn't been tested, only that they had achieved high performance ratings mainly by beating weaker players. Bu got draws from players like Aronian and Bacrot in the vital final games, which implies that either they couldn't beat him, or that they didn't think they could win without taking excessive risks. Thereby indicating a certain amount of respect. (This is the point I was trying to make about draws, which I admit could have been clearer).

Chinese players do indeed have difficulty getting invitations to supertournaments, but so do many players who aren't in the "magic circle"
(fortunately, organisers seem to be realising the important of introducing new blood). Didn't Zhang Zhong play at Wijk a couple of years ago?

An additional factor could be Government interference - are Chinese players allowed to play where they like, or are they subject to the type of restrictions that Soviet players used to face?

No... not scathing about Mark Crowther. I respect him quite a bit. My comment was based on the fact that it was mentioned as proof, but of course it was a cheating case and didn't apply in this discussion. It was apples and oranges.

I still disagree with your premise. You can't primarily beat weaker players and get high ratings. One loss (inevitable) and you're nosediving. Do you realize how many tournaments you'd have to play in to accomplish what you're suggesting? You have to beat strong players too or your rating will be stagant.

Certainly all of us without exception beat weaker players more frequently than stronger players. That says nothing at all... it is expected. However, Chinese GMs had to have competed at a high level to win as many accolades as they have.

You may remember that only a 4-0 collapse against Russia prevented them from winning the World Team tournament a couple of years ago. They have also beaten the U.S., France and Russia and goodwill matches. I could say more, but if you follow chess and read magazines, you know this.

Their players are beginning to travel more freely. When you have a trainer like GM Ye Jiangchuan, then results are expected. Hou Yifan continues to impress and there are several more young players coming up. The torrent of talent is too much to ignore. As the #3 ranked chess nation, China's future is as bright as anyone's.

No... not scathing about Mark Crowther. I respect him quite a bit. My comment was based on the fact that it was mentioned as proof, but of course it was a cheating case and didn't apply in this discussion. It was apples and oranges.

I still disagree with your premise. You can't primarily beat weaker players and get high ratings. One loss (inevitable) and you're nosediving. Do you realize how many tournaments you'd have to play in to accomplish what you're suggesting? You have to beat strong players too or your rating will be stagant.

Certainly all of us without exception beat weaker players more frequently than stronger players. That says nothing at all... it is expected. However, Chinese GMs had to have competed at a high level to win as many accolades as they have.

You may remember that only a 4-0 collapse against Russia prevented them from winning the World Team tournament a couple of years ago. They have also beaten the U.S., France and Russia and goodwill matches. I could say more, but if you follow chess and read magazines, you know this.

Their players are beginning to travel more freely. When you have a trainer like GM Ye Jiangchuan, then results are expected. Hou Yifan continues to impress and there are several more young players coming up. The torrent of talent is too much to ignore. As the #3 ranked chess nation, China's future is as bright as anyone's.

"I still disagree with your premise. You can't primarily beat weaker players and get high ratings."

It's not a premise, it's how the ELO system works. If you score 75% against your opponents, then your performance rating will be 200 points above their average, regardless of what your real strength is compared to theirs.

So if, for example, you regularly play opponents who are 150 points below you, and consistently score 75%, then your rating will be 50 points higher than it "should" be.

The number of tournaments does not affect this. Nor do losses, provided the overall score is the same.

It's quite common for players' ratings to be (legitimately) over-stated in this way, often in regions where the strongest players don't get many opportunities to meet opponents from other parts of the world. Latin America used to be a good example of this.

Anyway, as you say, we will no doubt hear more about the Chinese (their achievements, I mean, not their ratings!).

Anyway, we are talking about making it to 2650+, not merely having a rating that is 50 points over the stated rating. According to your formula, most of the players (outside the top 20) are about 50 points over-rated. Chinese are playing with the same rules as everybody else.

Top 20 players mostly play each other or those with similar ratings, so according to your theory none of them are over-rated. If that is the way the rating system is, maybe we should discuss the rating system instead of the Chinese been over-rated.

There ARE critics of the system and some believe there are far too many 2700s these days. Are we just getting a higher threshhold because of databases and computers or are we seeing rating inflation? Probably both.

Bu wins the Canadian Open beating Milov in the last round. Solid!

http://www.canchess.blogspot.com/

Bu wins the Canadian Open beating Milov in the last round. Solid!

http://www.canchess.blogspot.com/

"So if, for example, you regularly play opponents who are 150 points below you, and consistently score 75%, then your rating will be 50 points higher than it "should" be."--JR

"According to your formula, most of the players (outside the top 20) are about 50 points over-rated. Chinese are playing with the same rules as everybody else."--DS

The number of games played by each of the GMs in the Bu-Wang-Ni range over the past half-year in tournaments where their average opponent's rating was A) over 2600 B) 2550-2600 C) below 2550. All games included according to FIDE charts.

GM Name--A--B--C
Eljanov--11--24--5
Wang--0--18--16
Bacrot--15--3--5
Alekseev--26--9--0
Bu--9--13--36
Nisipeanu--14--11--15
Kasimdzhanov--9--4--8
Short--10--0--9
Almasi--0--20--6
Volokitin--30--8--9
Ni--0--9--42

It is legitimate for different GMs to have a rating of 2650 based on different patterns of opposition.

But to simplify for Daaim, let's try a different level of performance.

To be rated 2400, you can approximately:

Score 25% against 2600's, or
Score 50% against 2400's, or
Score 75% against 2200's.

Most people of 2400 rating would have played a mixture of opponents. But surely it is obvious that each of these % scores is not equally difficult for every player. Maybe if I trained fulltime for 5 years I could aspire to beat 2200 most of the time, but I think I'd never take points off a 2600.

I do not think anyone is saying the 3 Chinese players are over-rated, or that they could not score 50% against their own rating. The point is that they have tended to play lower rated opponents, so an interesting speculation can be raised.

The real strength of Topalov, Aronian, Carlsen, etc. has been questioned at various times on these blogs. People doubted whether Mamedyarov (another high elo by playing lower opponents guy) would cope at the elite level. Generally they have all proven themselves, just as Anand proved himself.

You all are now affirming my points. As I said, MOST players play lower-rated opponents. There are only 63 players rated 2650 or above. So, this stronger-weaker dichotomy is nothing new. It comes down to invitations and most of them go to European players... and Anand. Sasikiran is getting a few elite invites. Wang Yue and Bu have gotten a few, but they are far and few in between.

Yuriy,

Your stats confirm the contention that players over 2650 tend to play each other and all others play people lower. I don't see the logic in singling out Chinese players for playing weaker players.

gg,

I understand the numbers, but I don't think it is so simplistic. This thread IS about the Chinese being over-rated or not tested. Read the thread again. As long as those elite invitations are rolling in, they are almost certain to prove (or disprove) their strength.

Carlsen got constant invitations whether he did well or not. He still hasn't won anything major, but he'll keep getting invites. If Bu would have received the same number of invites to elite tournaments (when he was once the youngest GM), who knows where he could be. There are too many factors to control for to make these statistics generalizable.

Daaim,

"According to your formula, most of the players (outside the top 20) are about 50 points over-rated."

I don't see how this follows from what I posted above. All I did was give an example of how a player can achieve a high rating by good results against weaker players (which you were claiming was impossible - remember?).

"Top 20 players mostly play each other or those with similar ratings, so according to your theory none of them are over-rated."

On the contrary, closed groups are where unrepresentative ratings are most likely to occur (as in my Latin American example) - the Elo system assumes that all the players in the world regularly compete against all the other players, or at least a representative cross-section of them.

However, top 20 players often play other players in events such as Olympiads and the Bundesliga, or even tournaments like Dortmund which include players from outside the top 20. Also, players move in and out of the group with every new list.

gg,

"I do not think anyone is saying the 3 Chinese players are over-rated, or that they could not score 50% against their own rating. The point is that they have tended to play lower rated opponents, so an interesting speculation can be raised."

Exactly!

"Your stats confirm the contention that players over 2650 tend to play each other and all others play people lower. I don't see the logic in singling out Chinese players for playing weaker players."

I don't follow. Are you saying

a) the stats above do not show a difference in the caliber of opposition faced by Wang, Bu & Ni versus the caliber of opposition faced by faced by players who have the same rating

or

b) the players above do not all have a similar rating, one over 2650? (In fact, with exception of Eljanov who is ranked just before Bu, they all have a rating lower than Wang's and greater than or equal to Ni's).

James Raynard,

You gave two applied examples (Afromeev and Bloodgood) to cite your case. These were poor examples given that their ratings were manipulated illegally. The only other proof you give is theoretical (performance ratings).

I think someone else gave a 17-game analysis of Chinese players. Not a sizable sample at all. Again, I'll bet you that most players on the top list have beaten players 100-150 points lower 75% at least of the time. It's true across the board.

Why don't we agree that the rating system is flawed, but it is flawed for all players alike... not just the Chinese.

If there is no doubting that China is a powerful chess nation and their players are strong then this entire discussion is moot. Who cares about statistics. As Fischer would say, "All I care about is good moves."

Daaim,

"These were poor examples given that their ratings were manipulated illegally."

Which I mentioned at the time. The point was the inflated ratings, not the illegality.

"The only other proof you give is theoretical (performance ratings)."

That sounds as though it's meant to be a criticism.

First of all, being "theoretical" does not make a proof untrue. If that was the case, then mathematics would be impossible.

Secondly, you were claiming it was impossible to obtain an inflated rating by beating weaker players. To refute this proposition, only one counter-proof is required. Claiming I "only provided one proof" is a red herring.

If you insist on real-world examples, I can point you to Burma/Myanmar a few years ago, or Latin America (in the days before cheap flights), which I have mentioned several times in this discussion.

Yes, the rating system is flawed in this respect, and the tendency of the three Chinese players to mainly play opponents who are considerably weaker could have flattered their ratings. Other players with similar ratings play a wider range of opponents, so their ratings are not affected by this flaw. That is my argument. I'm not sure why you seem to be having so much difficulty grasping it.

"Who cares about statistics."

Well, someone on this thread accused a poster who raised this point of belittling Chinese players, not to mention being part of a Western bias against the rest of the world. So they were obviously important to him/her.

Certainly theoretically, the stats are what they are, but to use them without controlling for exogenous factors makes your argument misleading.

How can you eliminate the cause and effect in the examples you gave? Oh... "the point was inflated rating, not the illegality." Well... one caused the other. We are talking about people playing within acceptable guidelines, not Afromeev's or Bloodgood's hanky-panky.

Yes... you mentioned Latin America about three times, but no specific examples. Just "as in the case of Latin America." Latin America is a huge region including South America, Central America, parts of the Caribbean and North America (Mexico). Are you hinting at Henrique Mecking and Brazil in particular? Oh, and Myanmmar, another case of hanky-panky, right?

Yep... there is a Western bias in chess. No doubt about it. Have you ever been to a FIDE General Assembly?

You have no proof that similar players play a wider variety of opponents that the Chinese. These are merely your perceptions. Since this thread has started Bu Xiangzhi won the Canadian Open and Zhao Xue just won the Queen's tournament in Germany. Chinese players will keep winning while you keep theorizing why they're ratings are inflated.

Daaim,

I haven't been to a FIDE General Assembly, and I fail to see the relevance. However, the vast majority of FIDE members are from non-Western countries (which presumably means they are responsible for maintaining the likes of Campomanes and Ilyumzhinov in positions of power).

Getting back to the point, you stated that it was impossible to increase your rating by beating weaker players, and I have explained over and over again how it is possible. Whether it is done deliberately in order to cheat, or is just a coincidence, is irrelevant. It's still possible.

According to Yuriy Kleyner's post of July 15, 2007 23:53, the Chinese players in question played only 9 games against 2600+ players, 40 games against 2550-2660 players, and no less than 94 games against players below 2550. This was not the case for similarly-rated players in the same rating period.

Yet you say there is no proof. I can only assume that you are being deliberately obtuse, and I do not intend wasting any more time on your trolling.

James,

Anyone familiar with Daiim's many posts on this blog will recognize that he is definitely not being deliberately obtuse.

I've just realised I've made the colossal mistake of giving a sensible response to a troll (yes, I'm new to this game). So, with the indulgence of DD readers, I'd like to sign off with a more appropriate response.

"Latin America is a huge region"

Really? Due to my pro-Western biases, I had assumed it was the same size as the British Isles. And there's more than one country in it? Incredible!

Hang on - if it was so big, it might explain why chess players from there didn't get to play in many rated tournaments. Oops, almost got sensible there.

But at least they all speak Latin, right? (but only in front of foreigners - among themselves, they speak English, like everyone else does).

I can't wait to hear your next bombshell. Maybe you'll be telling us there's more than one species of bird. And as well as flying, some of them can actually swim!

(Actually, I'm not from the USA, I'm not 100% European, I speak three languages and the country I currently live in has one of the highest proportions of foreign residents in the world. But you seem to find facts confusing, so please feel free to continue making assumptions about me).

James,

Don't worry, Daaim is the guy who thought win, draw and loss were three equally likely outcomes in a chess game.

Also, funny how Daaim morphed from talking about 'European' to 'Western' as though these are synonyms.

Myanmar is not a case of hanky-panky?

"According to Yuriy Kleyner's post of July 15, 2007 23:53, the Chinese players in question played only 9 games against 2600+ players, 40 games against 2550-2660 players, and no less than 94 games against players below 2550."

James, that's not quite right. The games were placed in category based on average rating for their opposition in the tournament. I did it that way because my idea was to measure what kind of tournaments/competitions these guys play at. However, Gladiator's excellent original post demonstrates that if you look at individual games, the 3 Chinese players discussed above did tend to play players under 2600s a lot more than over 2600s and that virtually all of their ratings gain came as result of games against players ranked way below them.

Yuri, apologies for the misinterpretation.

I'm not sure whether anything was actually proved in Myanmar (if proof of such a thing is possible), but it looked highly suspicious. Didn't FIDE adjust the ratings of the players concerned?

I remember reading an article which claimed how they used Scheveningen-style tournaments to help things along, but I can't find it on Google (incidentally, the first two results I found were the government-controlled Myanmar Times and the almost as reliable Sam Sloan!)

James,

Here's your bombshell... or clusterbomb.

Troll? Nah. You admitted that you're the new one, right?

Organized chess is pro-Western and specifically pro-European, regardless of whether most of FIDE's members are non-Western and non-European. I have seen the stronger chess nations throw their weight around, BUT you're right in that the majority of the chess nations chose to re-elect Ilyumzhinov.

That is not say that is necessarily a bad thing (if it is in the process of evolution and not a standing order), but it is (and has been) a fact up to 2007. The bias comment was not directed at you any more than it is directed at players who refuse to believe that Chinese can play chess on an elite level, or there must be some anomoly in their ratings. There is a good reason... chess media does not traditionally cover these regions, so strong players are not know. Who knew about Wang Yue, Ni Hua, Wang Hao, or Hou Yifan? They came from nowhere and now everybody knows them. Who knew about Anand before he won World Junior? Yes... he had to prove his strength, but when the press found out about him, they looked at the quality of his games (and his calculating abilities) and the case was closed. They didn't go into gymnastics about his rating and how he was demolishing weaker players... which he had.

I'm sure you understand geography. All I was trying to get you to do was being more specific about the "Latin American" case and you still have not give a specific case.

James, I did not know your origin and it did not make a bit of difference in this discussion. I had already guessed you weren't Chinese, but beyond that it didn't matter. I could care less if you were American, Samoan, or Senegelese.

OK... you say Chinese "Big Three" have inflated ratings. You say they have played mostly weaker players. If you look at the average 2600-2650 Grandmaster, you'd find that also to be the case... especially if they are not getting the invitations. So let's say you're right in that.

gg,

You're bringing up an old argument which was resolved. There were two different theoretical arguments on the same thread. One was on the possible outcomes in a single chess game (three) and the other was probability given previous historical data. That much was understood in the end. Someone even concluded that chess was an unfair game due to white's perceived advantage. What about randomizing which color moves first? What about Fischer-Random? Well... blah, blah, blah... 200 posts later, some interesting points were made.

Yuriy,

Outside of the truly elite players, what players DO NOT gain most of their points playing weaker players? We have already discussed that there are other factors involved in WHY players play mostly weaker players (e.g., lack of invitations, distance, political red tape, etc.). Are we unfairly pointing at the Chinese when this is such a common trend?

Like I said, the Chinese will keep rising (and winning) while those you will continue to argue that they have inflated ratings. Beating the U.S., France and Russia in team matches is no small feat.

"Outside of the truly elite players, what players DO NOT gain most of their points playing weaker players?"

Such a statement if true would miss the point and support the accusation made against Bu, Wang and Ni. If one gains most points from facing lower-ranked competition, then the players who leap 40 points are the ones who are facing the lowest ranked more. The other players aren't facing the lower-ranked competition as often and aren't leap-frogging 40 points as result. But now let's see if the statement is true.

Above I already pointed out that players of rating similar to Wang, Bu & Ni do not FACE equally weak competition. Now let's see if players outside of elite GAIN most of their points against equally weak competition.

I started writing an answer to that question and found myself with what seemed to be a small dissertation. I think I have taken up enough space on this blog, so look for the answer here: http://olgasmove.livejournal.com/ and feel free to post your comments.
Shortly, the answer is no. The opposition the three Chinese GMs got most of their Great Leap Forward against is far from typical when compared to their peers.

One more interesting factoid. When I started posting in this thread, I assumed that most of the tournaments these 3 players faced the lower-ranked competition in were local and they perhaps indeed had difficulty traveling abroad, getting invitations from outside of China, etc. But here is where Bu, Ni & Wang actually have played over the past 6 months:

Wang--1 tournament in Russia, 2 in France & 1 in Philippines
Bu--1 tournament in Netherlands, 1 in Nigeria & 3 in Germany
Ni--1 tournament in Singapore, 1 in Russia, 1 in Nigeria, 2 in France & 1 in Philippines.

Whatever problems these guys might have, not being able to travel or not getting invited to a worldwide chess circuit is not one of them.

"Like I said, the Chinese will keep rising (and winning) while those you will continue to argue that they have inflated ratings."

That may very well be, but unfortunately those of us who aren't in possession of the fortuntelling seestones, have to rely on accomplished results and statistical analysis of completed games for indications of actual playing strengths.

"Beating the U.S., France and Russia in team matches is no small feat."

It's cute when somebody beats the drums of international warfare, but considering each of those teams beat China in the last Olympiad, and that results of such matches will differ greatly based on who represents each country, I have to ask what was the composition of US, France and Russia in these matches? Do you have a link to these results? Here is a just completed team match, in which China faced team Rest of the World, that was ranked on average 40 points below them and still lost by two points (and that was only because Wang Yue blew out Cheparinov). http://www.twic.co.uk/infoglueDeliverLive/article?contentId=711

Daaim,

"Here's your bombshell... or clusterbomb."

More like a damp squib, actually.

Sorry, but I don't have anything more specific about Latin America, as it was something mentioned in passing in an article. Perhaps someone who has a lot of spare time could go over cross-tables and games from the 70s and compare results and standard of play with published ratings?

It would not have affected players like Mecking, who mainly faced international opponents (when he was active), but more likely players around the IM level, who rarely got to play opponents from outside the region, due to the distance from the rest of the world and the size of the region. The numerous military dictatorships of the time probably didn't help much, either.

Something I can be more specific on is the USSR, which suffered from the same problem, but in reverse - the fact that the standard of play there was higher than anywhere else in the world was not accurately reflected in players' ratings. Again, the affected players weren't the top GMs who frequently played against their international peers, but those at the IM level and below.

Apart from super-GM showpieces in Moscow and Leningrad, like the Chigorin Memorial, international tournaments were very rare, and once the places for local GMs, comrades from "fraternal countries" and sympathetic Westerners had been allocated, there was only room for a couple of local non-GMs, who were no doubt closely scrutinised by Party officials before being allowed to take part. Internal travel was also strictly controlled, so there was no real possibility of playing in other parts of that vast country.

There were lots of untitled players rated around 2400, who had no problem achieving IM or even GM norms when they got the chance, but couldn't get titles as their norms expired before they got another chance to play in a rated tournament. They would of course have got a boost to their ratings from these performances, but only 10 or 20 points, when their real strength was more like 100 points higher than their published rating (as is obvious when you play through their games and compare them to their similarly-rated Western contemporaries).

In the Gorbachev era, these players were finally allowed to travel abroad and play in foreign tournaments (something that was previously allowed only for top GMs) and quickly demonstrated that they were indeed of GM strength.

Even before Gorbachev, a number of players who left the USSR achieved better results than their ratings would have indicated, at an age where players have usually stopped improving. The most famous example is Korchnoi, though he was already one of the world's top players. A much better example is Sosonko, who was 2420 when he emigrated in his late twenties; five years later he was sharing first place at Wijk with Geller and rated 2575 (at a time when there were less than a dozen players over 2600). However, it's not clear how much of this was due to under-rating and how much was due to other factors related to moving to the West, so I would prefer to concentrate on the later evidence.

I remember seeing an article which tried to calculate the ratings inflation among top players (does anyone have a link to this?) It concluded that the effect started in the mid-80s, with something like 30 points in 1986 alone, and had more or less died out by the mid-nineties. 1986 was when perestroika took off. Could the inflation have been caused by a sudden influx of players rated 2400-2450 who actually played 2500-2550? In which case, shouldn't it really be called a correction?

As for the three Chinese players, I have nothing against these players, or indeed the Chinese in general (as you appear to believe); if you read my first post in this thread again, you will notice that I was actually agreeing with much of what you said! Bu was clearly a very promising teenager, but, as with every other promising teenager, that doesn't count for anything until he delivers the goods.

Yuri,

Many thanks for your efforts. Trying to write an accurately researched and well thought-out post takes a lot of work, which may explain why so many people don't bother!

Thanks James for your detailed post!

I get the gist of your Latin American point. I thought you were talking about the Henrique Mecking controversy when he came from nowhere to rise in the world elite. Many people were stating that he had gotten there without being "tested" and his rating was not a true reflection. It was interesting, but of course he was found to be a great talent. It was tragic that he ran into personal problems and was not able to realize his potential. One other strong player to come out of Latin America was Columbia's Alonso Zapata. He was quite active on the world stage, but kept a steady rating without making a great leap. I believe you're right in a sense that there were not many regional opportunities, so these players had to travel.

I had a conversation with another player and we were talking about rating inflation. He said that perhaps players have merely gotten stronger (in the age of databases and computers). Of course this argument has been made before.

With 20+ players at 2700 and 12-13 year-old GMs (Bu was one as we know), I would say that I agree. There may be some inflation, but I would also argue that chess players (on average) have simply gotten stronger as they not only have more games and data to draw upon, but better and stronger tools to analyze the mass of data. However, I still like fingering through my 1953 Zurich book sometimes. ;-)

James, excellent post!

I think the article you are after might be 'FIDE Chess Rating Inflation' by Ron Edwards [ http://members.shaw.ca/redwards1/ ].

Looks a really good article to me.
He gives just 9 points inflation in 1986. And about 50 points inflation from mid-nineties to 1 April 2006 (the last date considered in the article). He gives a total of 118 points inflation from the end of 1985 to 1 April 2006.

I have always wondered what effect the FIDE decision in November 1986 (actioned in the January 1987 list) to increase all womens' (except Susan Polgar's) ratings by 100 points had on inflation. I remember seeing somewhere (FIDE forum?) that Ron didn't think that this was the reason, but didn't know what the reason was (though personally, I think it must have had at least some effect from, say, 1988). Your idea of an influx of under rated Soviet players getting out into the wider world might be the answer. Very interesting possibility.

It's interesting to convert some of the older ratings into 1 April 2006 terms:
Add 118 points to Fischer's peak 2785 rating of July 1972 and you get 2903(!).
Second highest is Kasparov's rating of 2800 in January 1990 - this converts to 2882.
Kasparov's official highest FIDE rating of 2851 in January 2000 converts to 2875.
Karpov's top rating comes out at 2843 (converted from 2725 as at January 1978).
Kramnik's top rating comes out at 2825 (converted from 2809 as at October 2001).

James' last post gives a very plausible, indeed, compelling, explanation for why FIDE ratings might have DEFLATED (NOT INFLATED) during the period he's talking about.

I'd never heard his explanation before, about the genesis and eventual exodus of a mass of underrated title-strength players from the former USSR. It certainly meshes with personal observation both before and after the former USSR opened its doors to the rest of the world.

The problem with his argument, though, is that letting a bunch of underrated players into a system would cause ratings to DEFLATE, not inflate. In other words, in James' scenario, the newly opened USSR would have been "exporting" deflation, as the outflow of underrated players coming out of there would take points away from everyone else.

This last isn't a point to argue about; it's a simple fact, like 1+1=2, as anyone can look up in the ratings literature.

There was a lengthy and very technical debate about rating inflation (both how to properly define it, and whether and to what extent it actually occurred) on USCF Forums several months ago. The two most active participants were Kevin Bachler and Todd Miller, who both claimed to have professional credentials as statisticians, as I recall. It was mostly about USCF rather than FIDE ratings, but readers who are USCF members can look it up.

A key point is that ratings in general can't go up because "chess players (on average) have simply gotten stronger"; the general advance of chess knowledge, databases, etc., has no effect on ratings in general. That's because any individual's ratings always reflect relative strength against the pool he or she plays against, never absolute strength.

Chris B, that was the link I was thinking of, though I seem to have mis-remembered the shape of the curve. Thanks for finding that!

Jon,

Your argument is of course quite correct.

This raises the question of where the inflationary element came from (assuming that there was inflation - any chance of a summary of the discussion for those of use who aren't USCF members?). I should have said that I'm only concerned with the top level, as analysed on the cited web page; things may very be different below, the top 50 in the world.

One possibility is from "over-rated" areas, like Latin America, but I doubt the number of players involved was anything like big enough.

Another possibility might be new players. Most of today's top 50 players still come from the former USSR and its satellites, so would have been affected by this "re-rating"; to illustrate, if the Karpov-Beljavsky generation faced opponents rated 2400 on the way up, who really should have been rated 2500, while players of the Kramnik-Topalov generation had similar results against similar players who were actually rated 2500, then the younger generation's ratings would have been 100 points higher, and this would have eventually distributed itself among their peers.

(Sorry if I haven't explained it very well - I've only just thought of it and am a bit pushed for time).

I fully agree with your final paragraph.

"A key point is that ratings in general can't go up because "chess players (on average) have simply gotten stronger"; the general advance of chess knowledge, databases, etc., has no effect on ratings in general. That's because any individual's ratings always reflect relative strength against the pool he or she plays against, never absolute strength."

That makes no sense to me as the rating spread/distribution should naturally *increase* when the highest level players improve relative to, say, a pre-database 1400.

Consider the position where one of two 1400 players who only ever play each other, SH. If one of them suddenly gets 200 pts stronger, he will become 1500 in time and his opponent 1300. If both get 200 points stronger, they will both remain at 1400. This in microcosm is the point; there is no objective gold standard by whch ratings can be measured.

Releasing underrated 2400s into the ecosystem shouldn't have any flationary effect either de- or in- either, since when 2400s play 2500s it is a zero-sum game.

What would have an effect is releasing a whole bunch of underrated players below 2400, since when they outscored their expected score against 2400+s they would be going up with a k factor of 15 (that is 15 ELO points for every point they outscore their expected score by), whereas the 2400+s would be going down only at a rate of 10 per point less than their expected score. So the overall average rating would inflate as you would have more points in the system. This is of course exactly what happens when juniors come into the system in general, and believe the whole k-factor shift at 2400 was suppoed to offset this (in the days of course when the rating floor was 2200). But I suspect that this can't be a big enough factor to be an explanation. Not that I have a better one.

It may of course be that ratings at the top end have gone up because the gap between Kasparov and the players rated say 2200 genuinely has increased. Whether that would increase the average rating depends I suppose on what happens to players driven ut the bottom of the system, and I don't know the answer. What does happen to you if your ELO falls beneath the minimum. Do you just disappear, taking the points you just lost out of the system with you?

Instead of two members, consider an infinitely-sized population.

Suppose top 1% harvest 100pts in database bonus, though non-professionals like 1400s do not.

Does the distributional spread increase? *All else being equal,* it must. The 1400 rating isn't completely arbitrary; he must still win x% from a 1100, a 1100 x% from 800. Therefore the 100 points tends to squeeze out the high end.

Yes, that must be right. If the strong get stronger relative to the weak, the range will grow and hence so will the grades at the top. But that won't make the average grade inflate, in fact if professionals grow more better than amateurs you'd expect the average to deflate, I'd have thought, as 2500/2300 matches go more the way of the former than you'd expect.

Maybe we need to agree first of all what phenomenon it is we're trying to explain.

Perhaps we ought to be considering what

There have been other discussions of FIDE ratings inflation on Dirt, and like this one, the people most interested in it are concerned with the proliferation of very high ratings (GMs, top 50, and the like).

The explanation I've given, in line with rdh's remarks immediately above, is that higher ratings for the top 50 on the one hand, and "rating inflation" on the other, are entirely distinct phenomena.

I've explained the big climb of the top 50 as resulting from a progressively larger pool of rated players. The larger the population, the greater the difference between the extremes and the mean. Individuals 5 standard deviations stronger than the mean would amount to at most a handful in a (normally distributed) universe of 10,000 rated players; but in a universe of 1 million rated players, you'd expect to find hundreds of them, and even some 6- and 7-sigma outliers. That's my explanation for why there are as many 2700s today as there were 2600s in the 1980s. It's not inflation, but something else.

As for inflation: By strict analogy with economics, "inflation" refers to the GENERAL price level. For the case at hand, then, ratings inflation means a higher GENERAL or overall average rating level, as opposed to just the top of the food chain. As we've already agreed, this can never have any connection with the advance of chess knowledge (hence average absolute strength) over time.

It seems to me that changes in FIDE rating eligibility make the whole question of general rating inflation moot -- that is, impossible to measure in any meaningful way.

In a mechanical, mathematical sense, obviously it must be that the simple average rating of all FIDE-rated players has declined over time, and declined big-time, simply because FIDE keeps lowering the rating floor. But this shouldn't be considered rating deflation or inflation. (Aside to rdh: I'm pretty sure the FIDE rating floor systemm works the same as in the USCF and all other such systems. You never lose your rating. If it falls below the floow, it simply remains at the floor. Floors are the most obvious potential cause of rating inflation: players in general being overrated. Someone at their floor is overrated, by definition. Others who play that person will win more often than implied by ratings-based probabilities; so the floored player's "excess" points thereby leak out to the rest of the rated population, via their opponents.)

Yep... China just beat Russia... 52.5-47.5. I suppose naysayers claiming Chinese are overrated and cannot beat players over 2650 will now admit to their folly. The Chinese men won the last round +3.

triple't'disaster(tomashevsky,timofeev and tairova)for team russia.tomasevsky and tairova were debutants,bad surpise from timofeev the best player of'world'team at world vs china 2006.probably that result was the reason timofeev played for the russian squad,since his results in 2006,2007 are very poor for his talent(and rating)the first 3 players for russia'b'males scored a nice +5(jakovenko's result was still an undeperformance at 50%)tairova buried the female squad since the others girls scored a+1 combined.the chinesse squads looked more compact and scheveningen experienced(they played twice against'world'teams plus against UK and russia within 1 year),the young chinesse male squad had no rooky.still alekseev was head and shoulders above the rest of the field(didn't loose a game in dormund and in this tourney

triple't'disaster(tomashevsky,timofeev and tairova)for team russia.tomasevsky and tairova were debutants,bad surpise from timofeev the best player of'world'team at world vs china 2006.probably that result was the reason timofeev played for the russian squad,since his results in 2006,2007 are very poor for his talent(and rating)the first 3 players for russia'b'males scored a nice +5(jakovenko's result was still an undeperformance at 50%)tairova buried the female squad since the others girls scored a+1 combined.the chinesse squads looked more compact and scheveningen experienced(they played twice against'world'teams plus against UK and russia within 1 year),the young chinesse male squad had no rooky.still alekseev was head and shoulders above the rest of the field(didn't loose a game in dormund and in this tourney

Yuriy, gg and James Raynard,

After China beat Russia (overall, men and women), we heard nothing from you... the naysayers who stated that Chinese players could not beat players over 2650+ and were over-rated. Well... all the Russian players were rated at least 2650. Where are your tables and statistics to explain this result?

I'm surprised Mig didn't post on this major event.

Daaim,

Well speaking for myself I put out such anti-Chinese words as:

"I do not think anyone is saying the 3 Chinese players are over-rated, or that they could not score 50% against their own rating. The point is that they have tended to play lower rated opponents, so an interesting speculation can be raised."

Do you have anything concrete to complain about?

You like to attribute some viewpoint to others then ignore what they actually say.

Daaim,

Could you please provide a quotation in which I said that Chinese players could not beat players over 2650?

The reason why I didn't post anything was simply because I had rather more pressing concerns last week than following chess results (unfortunately such things do exist). I suspect something similar would explain Mig's not mentioning it.

Looking at the report on Chessbase, it appears the Russian men were leading until they collapsed in the last round (a curious reversal of the Chinese collapse in the World Team Championships last year). So a good result for the Chinese men, considering the Russians were higher graded and had home advantage.

The Russian women never managed to recover from their disastrous second round, although they did put up a reasonable fight in the rest of the match. Again a good result for the Chinese, who didn't field their strongest team (no Hou Yifan).

James Raynard and gg,

A lot of time was invested in proving that Chinese players got to where they are by playing players much weaker. What is the implication? That they are not as strong as their rating indicates. There was even a sub-debate on rating inflation.

Yuriy was actually the one who stated these theories more forcefully than both of you by providing stats and doing an analysis of their competition... flawed and non-generalizable. Everyone seemed to agree with his analysis.

Plainly, the Chinese are a strong lot and their ratings (however they got there) are indicative that the are a top three federation in the world... by federation rating, by Olympiad results, by beating other federations (US, Russia twice, France, next UK), by playing ambitious chess, by winning tournaments.

There was no need whatsoever to cast a slither of doubt on their strength and that is what the entire conversation was about. China will rule chess within 10 years as they have almost an unlimited supply of talent... and they are more focused than any other federation, bar none!

I remember when Ni Hua was a 2568 untitled player and basically skipped the IM title. There are several more players like that in China. Wang Hao is another example. I'm not sure he had a title when he became a GM. When Zhang Zhong came to the U.S. with the Chinese delegation, he had a 2607 rating and I had never heard of him! He has switched his federation to Singapore, but China is still strong.

James Raynard, you did praise the Chinese, but in your posts, you harbored doubts on their strength. Last year, people were saying that the Chinese shouldn't have won the Olympiad silver because they beat up on "weakies" and Russia (who did not medal) had a better record and played closer matches against stronger teams. Perhaps the thought of China playing weakies is a pervasive perception. I believe these arguments about China will soon be a thing of the past as China continues to win tournaments and matches.

Daaim,

All I said was that the Chinese had not been tested against the strongest opposition, which was true.

The reason China won the silver medal was because they scored more points than every other team (except Armenia). I have never claimed otherwise - in fact, I have never even commented on the Olympiad, so I'm not sure where that came from.

In fact, there was some discussion of the Olympiad, in which I stated the results supported both views.

However, it's a long time since this thread had anything that was of interest to me, and I have better things to do than respond to accusations of saying things I didn't say, or of believing things that you've made up.

Goodbye.

James,

The Olympiad comments were not directed at you. I was referring to the general climate last year and common perception that China won the silver on some type of a fluke, were overrated and had not been tested in the tournament. I was a part of those conversations. Some of the same arguments were made on this thread. Well... they'll diminish gradually once we realize that chess has become more universal than ever.