chessgames.com

TOURNAMENT STANDINGS
King's Tournament

Magnus Carlsen7.5/10(+5 -0 =5)[view games]
Boris Gelfand5.5/10(+3 -2 =5)[view games]
Teimour Radjabov5.5/10(+3 -2 =5)[view games]
Ruslan Ponomariov4.5/10(+2 -3 =5)[view games]
Liviu Dieter Nisipeanu4/10(+2 -4 =4)[view games]
Wang Yue3/10(+0 -4 =6)[view games]

 page 1 of 2; games 1-25 of 30  PGN Download
Game  ResultMoves Year Event/LocaleOpening
1. Carlsen vs Ponomariov ½-½31 2010 King's TournamentE06 Catalan, Closed, 5.Nf3
2. Nisipeanu vs Radjabov 1-039 2010 King's TournamentB33 Sicilian
3. Gelfand vs Wang Yue 1-082 2010 King's TournamentD10 Queen's Gambit Declined Slav
4. Radjabov vs Carlsen ½-½74 2010 King's TournamentB25 Sicilian, Closed
5. Gelfand vs Nisipeanu ½-½32 2010 King's TournamentD20 Queen's Gambit Accepted
6. Wang Yue vs Ponomariov ½-½40 2010 King's TournamentE06 Catalan, Closed, 5.Nf3
7. Ponomariov vs Radjabov 0-148 2010 King's TournamentE81 King's Indian, Samisch
8. Carlsen vs Gelfand ½-½40 2010 King's TournamentD43 Queen's Gambit Declined Semi-Slav
9. Nisipeanu vs Wang Yue ½-½31 2010 King's TournamentC42 Petrov Defense
10. Radjabov vs Gelfand 1-038 2010 King's TournamentC42 Petrov Defense
11. Ponomariov vs Nisipeanu 1-036 2010 King's TournamentE10 Queen's Pawn Game
12. Carlsen vs Wang Yue 1-054 2010 King's TournamentC36 King's Gambit Accepted, Abbazia Defense
13. Nisipeanu vs Carlsen 0-131 2010 King's TournamentB76 Sicilian, Dragon, Yugoslav Attack
14. Wang Yue vs Radjabov ½-½41 2010 King's TournamentA15 English
15. Gelfand vs Ponomariov 1-038 2010 King's TournamentD38 Queen's Gambit Declined, Ragozin Variation
16. Radjabov vs Nisipeanu ½-½30 2010 King's TournamentD27 Queen's Gambit Accepted, Classical
17. Ponomariov vs Carlsen 0-150 2010 King's TournamentE81 King's Indian, Samisch
18. Wang Yue vs Gelfand  ½-½31 2010 King's TournamentD39 Queen's Gambit Declined, Ragozin, Vienna Variation
19. Ponomariov vs Wang Yue ½-½63 2010 King's TournamentD10 Queen's Gambit Declined Slav
20. Nisipeanu vs Gelfand 0-140 2010 King's TournamentC42 Petrov Defense
21. Carlsen vs Radjabov 1-064 2010 King's TournamentB30 Sicilian
22. Wang Yue vs Nisipeanu 0-166 2010 King's TournamentE64 King's Indian, Fianchetto, Yugoslav System
23. Gelfand vs Carlsen ½-½33 2010 King's TournamentD37 Queen's Gambit Declined
24. Radjabov vs Ponomariov  ½-½30 2010 King's TournamentC55 Two Knights Defense
25. Radjabov vs Wang Yue  ½-½32 2010 King's TournamentC42 Petrov Defense
 page 1 of 2; games 1-25 of 30  PGN Download
  REFINE SEARCH:   White wins (1-0) | Black wins (0-1) | Draws (1/2-1/2)  
 

Kibitzer's Corner
< Earlier Kibitzing  · PAGE 30 OF 30 ·  Later Kibitzing>
Jul-01-10  frogbert: <The types of chess they play is so different, that it wouldn't be much different if one would've introduced a new piece or allowed the bishop to jump over one piece diagonally, given that it didn't move more than 2 (or 3) squares, now wouldn't it ?>

haha. the finer points about changes in chess relevant for the world elite players (from where you got that quote and most of your "points"), is quite irrelevant for two players who are already divided by an ocean of chess understanding.

if the "real" difference turns out to be equivalent to 350 or 450 points (instead of 400) based on 4-6 games, then nobody really cares, and the 2600-player will crush the 2200 player convincingly regardless, now as he did in 1990.

however, if someone would suggest that a 2010 player rated 2826 is nearly as good as a 1990 player rated 2800, then 1000 chess fans will get mild heart attacks, their faces will go red, and chess forums will be spammed with angry posts. :o)

<therefore> my points from chessninja are valid for elite chess, but not so relevant in your clever example.

Jul-01-10  Blunderdome: <How would you estimate the expected outcome of a 20 games match between, say, an 2100 player from Norway and an 2600 player from Russia?>

Well, frog could just play a 20-game match again Sergey Ionov, Alexander Rustemov, or Alexander Volzhin, and tell you what the result of that match was ;)

Jul-01-10  frogbert: maybe i'd get a draw or two if i were lucky? :o)
Jul-01-10  Kramer: <i suggest you stick to the topic of debate in the future and refrain from telling other people how they should address their friends.>

I thought maybe that could be the case, that you knew him from before... and that you thus didn't need to be as polite as with a complete stranger. I thought of that afterwards, though.

I wrote what I wrote, because your response would have been rude, if you did not know him.

Or, rather, it appeared rude to me.

Also, if you take the pleasure of being as direct as you are to me here, i will not hesitate to be equally direct to you in the future. You stick to the topic, I'll stick to the topic. ;)

Jul-01-10  metatron2: <frog: the finer points about changes in chess relevant for the world elite players (from where you got that quote and most of your "points"), is quite irrelevant for two players who are already divided by an ocean of chess understanding.>

That doesn't explain why you consider two ratings numbers that represent results against two <completely different> pools of players, to be perfectly comparable, while you consider two rating numbers to be totally incomparable, if their respective pools are different because of a time gap, instead of a rating+geographical gap.

As for the rest of the points: What if you look at 2250 vs 2500 (or even 2300 vs 2500) instead of 2200 vs 2600 players? The difference in chess understanding is not that huge any more, yet my points are still relevant since it is still amateur vs pro/semi-pro.

Why do you consider the rating system to be reliable in such cases, while you consider it to be unreliable if similar differences are generated over time ?

<if the "real" difference turns out to be equivalent to 350 or 450 points (instead of 400) based on 4-6 games, then nobody really cares, and the 2600-player will crush the 2200 player convincingly regardless, now as he did in 1990.>

So are you actually saying that you don't find the rating system reliable in case of 400 points gap ? what about 300 or 250 points gap ?

Also recall that I was talking about 20 games match (and not 4-6 games).

Jul-01-10  frogbert: kramer, i have no interest in discussing personal issues or my or other kibitzers' behaviour.

i'll make one small clarification, though. in the following:

<adjusting current ratings up or down (or painting them in new colours) doesn't create any anchor to skills - except possibly in some people's heads.>

i did not mean to refer to "zarg's head" (or mind) [an impression you possibly could've got], i just expressed that i wouldn't be surprised if some people would be guilty of the mentioned misconception.

with that sorted out, i'll just assure you that as long as you refrain from personal comments about me or how you perceive my behaviour, i'll make absolutely no such comments related to you and your presence on cg.com.

but please be direct: if you think something i write makes no sense or is logically or factually wrong, don't hesitate to say just that, with no sugar coating. however, i'd appreciate an explanation of why you think i'm wrong, of course. :o)

Jul-01-10  frogbert: <That doesn't explain why you consider two ratings numbers that represent results against two <completely different> pools of players, to be perfectly comparable, >

i didn't claim that - you have added the prerequisite that their numbers represent results against "two <completely different> pools of players" in retrospect - it does in no way follow directly from your original question to me.

norwegian players play a lot abroad in international tournaments, thereby ensuring that their fide ratings are very well adjusted to the european sub-pool (we're talking thousands of recent games), while a 2600 gm from russia (or most countries, actually - possibly except china and similar distant countries) would also play the majority of his games abroad or against other players that regularly play abroad.

in short, the interaction between "regional subpools" today can be <demonstrated> to yield very comparable ratings (did you see the calculations i posted on chessninja, showing that us 2400+ fide ratings were basically in perfect harmony with european 2400+ ratings?). a similar demonstration can <not> be made with players divided by time instead of space.

also, i suppose you know my "rule of thumb" from previous discussions: rating comparisons over time make reasonable sense up to ca. 5 years (although gradually degrading in accuracy), while i advice increasingly bigger grains of salt to be applied when the time gap goes towards 10 years. after roughly 10 years i think the number and size of changed factors are making it very questionable to assume too much on the basis of ratings alone.

hence, i do "allow" a certain time gap - actually quite a big one, i think - before i say "stop". and note: i consider the "gap" across time to be of a fundamentally different type/nature than any geographical gap in our modern, globally interconnected today.

Jul-01-10  frogbert: <So are you actually saying that you don't find the rating system reliable in case of 400 points gap ? >

i find it realiable; i'm perfectly convinced that the higher rated player has used to have much better results than the lower rated one, making me think the rating favourite will win any match of 5+ games very convincingly. :o)

an exact prediction of a <match outcome> between two specific players can never be made based on the ratings alone, though. no matter how close the ratings are to each other.

first, any player's rating mostly reflect tournament play, not match play. secondly, between a specific pair of two players, there are a number of additional factors to be taken into account. and finally, ratings aren't first and foremost a predictor, but a <descriptor>. it's a misunderstanding to think that ratings give any guarantees about the future - they are simply a (usually pretty good) summary of <past> results.

the latter should be part of the rating abc, but let me just give one little example to illustrate this point.

two players, a and b, play a match. both are rated 2500. one year prior to the match, player a was rated 2400 and player b was rated 2600. the current rating number says absolutely nothing about current form, recent rating development, the players' ages, health and so on. but they describe the two players' past results as good as the rating system possibly can.

what is the "expected" outcome of a 10 game match? technically, for the rating adjustment formulas, both players need to score 50% to keep their rating - but the rating system does not <predict> that this will be the outcome, it merely requires a 50% score for each to retain their ratings. no bells and whistles go off if player a wins 6,5-3,5 - the rating system is perfectly happy and simply adjusts the two players' ratings accordingly, stating that it's got a new and updated number describing the players' accumulated results. that's what there is to it, a summary of past results.

Jul-01-10  metatron2: <you have added the prerequisite that their numbers represent results against "two <completely different> pools of players" in retrospect - it does in no way follow directly from your original question to me.>

Well, that was the main idea behind my original question, and appeared as the first item in my players comparison list in the post you replied too.

<norwegian players play a lot abroad in international tournaments, thereby ensuring that their fide ratings are very well adjusted to the european sub-pool (we're talking thousands of recent games), while a 2600 gm from russia (or most countries, actually - possibly except china and similar distant countries) would also play the majority of his games abroad or against other players that regularly play abroad.>

I'm not sure how many international tourneys 2200 Norwegian players actually play, but even if they played a lot of them, how much overlap they really have with the pools 2500-2600 players play? Even when they play in the same tourneys with the GMs, they usually meet quite different pools of players in most of the rounds. Also they usually play in different leagues (or at least different boards in their league teams), that further separates their pools.

<i consider the "gap" across time to be of a fundamentally different type/nature than any geographical gap in our modern, globally interconnected today.>

That should be: geographical <and rating> gap. And I still didn't understand why consider those two to be "fundamentally different".

<i suppose you know my "rule of thumb" from previous discussions: rating comparisons over time make reasonable sense up to ca. 5 years>

Of course I know it. That's what I'm actually challenging here frog ;)

My question remains:

Here you have 2 players, playing in 2 different pools of players, having various different types of chess properties, and yet you assume that the rating system is strong enough to cover those big gaps. How come ?

Jul-01-10  frogbert: i've already answered:

1) geographical "gaps" are overcome on a daily basis: it's called travelling. time travelling is fundamentally different.

2) repetition: <in short, the interaction between "regional subpools" today can be <demonstrated> to yield very comparable ratings (did you see the calculations i posted on chessninja, showing that us 2400+ fide ratings were basically in perfect harmony with european 2400+ ratings?). a similar demonstration can <not> be made with players divided by time instead of space.>

hence, i can give you empirical evidence that geographical gaps represent no big challenge (today). you can only speculate about time gaps.

Jul-01-10  metatron2: <i've already answered>

No you did not, because as I said, its not just "geographical gaps", but "geographical <and rating> gaps". And you didn't answer that at all.

And regarding geographical gaps only:

How would you estimate the percent of games an average 2550 Russian player plays outside Russia? And what about the percentage of games an 2200-2300 Norwegian player plays outside his nation? I'd say less then 40%.

Can you give an estimated average overlap between the 2 pools of players those 2 types players (with their rating gap!) are expected to meet? What should be? 5%, 10%, 30%, 50%, or maybe more?

My guess would be around the 5%.

Jul-01-10  frogbert: as i already told you: it can be demonstrated that the interaction is big enough for present day players. simple empirical methods are sufficient to do that: players from (very) different geographical areas score basically as "expected" from their fide ratings also when they play abroad, and when foreigners play in "their" country. the only notable exception i can think of is young 2400 and 2500-rated chinese (and other east-asian) players that seem to have too little interaction with the rest of the (current) pool, and they appear to become underrated as a consequence.

the general interaction between russian (2600-)players and western-european players is certainly as big as that between us and european players, and even in the latter case, where the contact is basically limited to less than 100 players going each way, calculations showed amazing coherence.

the strength of transitive relationships are never 100% in chess, but what you can infer about a and c if a plays b and b plays c is <radically more> if the games take place within months or a year, as compared to a scenario where a plays b, 10 years passes, and b plays c.

gaps in space (and rating) are covered quite fast and reliably by transitive relationships in our modern world - again, as can be <demonstrated> by looking at how the chess travellers do home and abroad. i don't have to guess about sub-pool overlaps in order to show that.

but while we have chess travellers in rather decent numbers, the time travellers are few. if gelfand beat karpov convincingly in 2002, it wouldn't say much about how gelfand would've done against korchnoi in 1978. introducing numerous links on the path between the two games/matches is just an alternative source of error. and then we haven't even started to talk about the changes to the environment in which chess is being competed at.

Jul-01-10  metatron2: <the general interaction between russian (2600-)players and western-european players is certainly as big as that between us and european players>

I'm sorry frog, but talking about general interactions between 2600- players, doesn't really answer my question, regarding the estimated (average) pools-overlap of two specific players: an 2250 Norwagian and 2550 Russian.

Here is my question again:

"Can you give an estimated average overlap between the 2 pools of players those 2 types of players (with their rating gap!) are expected to meet? What should it be? 5%, 10%, 30%, 50%, or maybe more? My guess would be around the 5%."

<frog: and then we haven't even started to talk about the changes to the environment in which chess is being competed at.>

Actually I talked about those competing-environment changes right from the start..

<gaps in space (and rating) are covered quite fast and reliably by transitive relationships in our modern world>

OK, Now we are getting somewhere:

If I understand correctly, What you are saying here is that the 2 players can play against <different pools>, as long as you assume that transitivity is kept between the players they play against (from the different pools). And you assume that if we have enough interactions between the pools, the transitivity property will indeed be kept.

Right?

<it can be demonstrated that the interaction is big enough for present day players. simple empirical methods are sufficient to do that: players from (very) different geographical areas score basically as "expected" from their fide ratings also when they play abroad, and when foreigners play in "their" country.>

I am sure they score around their expected ratings. That's my <whole point>.

You claim that because the pools change over time, and because of competing environmental changes over time, there is no meaning at all in trying to compare ratings separated by time, or talk about "rating inflation" with regard to chess skills.

But if we can show that the old pool that changed over time, had enough "interactions" with the new one, there is no reason for losing the transitivity property between those two pools.

And we also have big competing-environment-differences between players within the current pool, and yet those differences do not invalidate rating comparison between such players, and so there is no reason for such changes to invalidate comparisons of ratings separated by time.

What I wanted to show, is that you cannot completely rule out such comparison and the whole concept of rating inflation with your arguments, since we can find similar symptoms to those you mention within players in the current pool.

I'm not saying that the rating inflation problem is solvable. I'm just saying that you canít rule out the entire concept.

Now I'm going to sleep. To be continued tomorrow (today actually..).

Jul-01-10
Premium Chessgames Member
  Bobwhoosta: <metatron2>

I believe you're missing <frog's> point here.

His claim is that the FIDE ratings of current players is relatively accurate in describing past performance relative to the field- and therefore somewhat reliable in predicting match results- because of a simple principle... (enter pure Whoosta comments):

When a player enters a tournament, every player in that tournament is carrying direct information from a series of players they have (in most cases) played recently. When Player A plays Player B, it may be they have not played in the same pools recently. Possibly never. However, they have played people who have played people who have played people... And so on. By the time you iterate this information to the third or fourth step, you have branched out significantly into the world population.

Example: Our Norwegian player, an amateur, plays in a local tournament. In this tournament there are two NM, one IM, and one GM, along with a significant number of amateurs of varying strength. This is not a stretch of an assumption. The NM, IM's, and GM bring into the tournament information from every player they have competed against, and the amateurs the same. Whether the amateur even has the opportunity to play the GM, he will most likely play someone who will play someone who will play the GM. Therefore, the information latent in the GM's rating, being derived from a world population, will influence the tournament. This may seem to be a small influence, and yet it is continuous, offering many little corrections to keep the rating system fairly accurate. And it is impossible to assume it is merely the GM who is playing outside the Norweigan's "pool" of players, because there are many amateurs who travel, and the Norwegian will encounter enough 2400 rated players (who play against enough 2600 rated players) to ensure any discrepancy in ratings between separate locations remains small.

The difference between geography and time is not a quantitative difference, rather a qualitative difference. 10 years in the past cannot be represented in any tournament today, unless of course you have a GM who quit playing 10 years ago, and decided to play in the tournament you're in. As mentioned before, this "GM" effect would be decidedly small, and must happen numerous times in order to have a cumulative correcting effect on the population.

How many GM's IM's, and amateurs do you know of who haven't touched the game in 10 years, and are now entering the ratings pool?? Therefore, you can assert that- within simple limits- the length of time between the ratings of two players will have a negative effect on the accuracy in predicting a match outcome.

Jul-02-10  metatron2: <Bobwhoosta: <metatron2> I believe you're missing <frog's> point here.>

Not at all, I completely understood his point, and you've just described in detail the following conclusion from my previous post:

<What you [frog] are saying here is that the 2 players can play against <different pools>, as long as you assume that transitivity is kept between the players they play against (from the different pools). And you assume that if we have enough interactions between the pools, the transitivity property will indeed be kept.>

As you could see from that post, my main point was this:

<if we can show that the old pool that changed over time, had enough "interactions" with the new one, there is no reason for losing the transitivity property between those two pools.>

And regarding such "interactions" between pools that are separated over time, I think your analysis were wrong here:

<Bobwhoosta: 10 years in the past cannot be represented in any tournament today, unless of course you have a GM who quit playing 10 years ago, and decided to play in the tournament you're in. As mentioned before, this "GM" effect would be decidedly small, and must happen numerous times in order to have a cumulative correcting effect on the population.>

The players do <not> need to quit during the separating time period, in fact quitting would probably make them rusty and degrade their level of play. What you actually need, is to have <enough> players remaining in the pool <without> significantly changing their level of play.

Take 2 pools separated by 2 yrs period: if say, 30% of the players in year Y, remain active in the pool without significantly changing their level of play until yr Y+2, then we have enough information "carried" from the old pool to the new one, and similarly to your example, this information is delivered to the rest of the players in the new pool, in a continuous way.

And if that is true, then there is no reason why the same won't happen between the pools from yrs Y+2 and Y+4 and so on, keeping the rating relevant between two pools with 10 yrs gap, <unless> at some point there are not enough players remaining in the pool without significantly changing their level of play.

So let's try to evaluate the probability for such "static" players to remain active in the pool:

From my experience, the larger part of the amateurs players, stop making progress around their early 20's (or at least significantly reduce their progress rate), mainly because they start studying, working, building families, etc. and don't have much time to work on their chess (playing online blitz doesn't really improve your chess ;) ). As for the pro/semi-pros, I think mostly play on similar level between the age of 30-45, while many of them reach plateau long before that (simply because they worked hard and reached they full potential early). Finally, most amateurs can probably keep their level of play well into their 50's, because they don't have such tough opposition, that requires high energy levels from them, etc.

So here is what we get from all that:

Most of the (none beginners) amateurs, are expected to remain around the same level during those 2 yrs period, if they are in the [ 20+ to 50+ ] age range, while the pros are expected to be "static" if they are in the 30-45 age range, with many of them becoming "static" while being in the [20-30] age range as well.

So I'd say that more then 50% of the pool is expected to "carry" the old pool's information, which should be more then enough.

---

So to conclude:

There are various factors that can cause rating distortions (or "rating inflation") over time. The changing players in the pool is one of them, but because of the <continuous way> the pool is changing, it is by no means something that invalidates rating comparison from different periods, just one more thing that can "cause problems". And trying to compare rating from different periods of time, is by no means "meaningless" (as frog says).

It probably will require oceans of work even just to evaluate whether there exists some mathematical model that can really track rating inflation using formulas, but the whole <concept> of chess-skills based rating inflation, is a totally <valid> one.

Jul-02-10
Premium Chessgames Member
  Bobwhoosta: <metatron2>

Good point, I'll think about that a bit more...

Jul-02-10  Chessforeva: 3D games: http://chessforeva.appspot.com/C0_p...
Jul-02-10  frogbert: <And trying to compare rating from different periods of time, is by no means "meaningless" (as frog says).>

i don't say, as i think you know, that comparing ratings from different "periods of time" is meaningless regardless. if the changes aren't too big

1) in the pool sizes

2) in the pool content

3) in terms of external changes in the environment of competitive chess

then such a comparison makes sense to some degree. comparing ratings from 2010 and 2009 and 2008 i find quite unproblematic.

however, i say that with the huge changes in the fide rating pool over the past 25 years, multiple changes to the lower threshold for fide ratings, <plus> the numerous changes to top level chess (time controls, computer tools and internet, etc.) then

a) i don't recommend comparisons spanning more than 10 years (less changes would've given a different estimate - the estimate is linked to the amount of change)

b) comparisons become increasingly meaningless (as a tool for comparing skills) as the time span widens

the only thing it is <possible> to monitor and measure, is <nominal rating inflation/deflation>, since ratings are based on results and results only. if meta wants to believe something else, he probably also needs to accept that carlsen currently is less than 50 points from kasparov's peak rating - with rating as a "measure of absolute skills". [disclaimer: i don't advocate this use of the fide ratings.]

i'm not sure if meta is ready to bite that bullet.

Jul-02-10  frogbert: <I'm sorry frog, but talking about general interactions between 2600- players, doesn't really answer my question>

i already said that i <do not need to answer that question> in order to say something meaningful about whether the sub-pools are interconnected enough or not. i quote:

<gaps in space (and rating) are covered quite fast and reliably by transitive relationships in our modern world - again, as can be <<<demonstrated>>> by looking at how the chess travellers do home and abroad. <<<i don't have to guess>>> about sub-pool overlaps in order to show that.>

so, if you are sorry, then i am sorry too, that you won't accept the answer i'm willing to give. for contemporary players divided by geographical distance, the assumption about "enough transitivity" can be <tested and demonstrated> to hold up convincingly, while your assumption regarding players divided by 10-15+ years can not possibly be tested in any way.

<that> is my big point that you don't seem to consider. :o)

Jul-02-10  frogbert: < there is no meaning at all in trying to [...] talk about "rating inflation" with regard to chess skills.>

one can talk all one wants and speculate that one gets more or less "skills" per rating point as compared to before, but i can't think of any method of making near-scientifical tests for whatever hypothesis one has. and i haven't seen any serious suggestion from anybody else, either. please feel free to suggest something that can be tested.

the situation is however <different> when it comes to <geographical> distances and geographical (or ratinglevel) sub-pools.

Jul-05-10  metatron2: <frog: comparisons become increasingly meaningless (as a tool for comparing skills) as the time span widens>

Well there is a big difference between saying:

(a) "I don't recommend directly comparing rating from different periods of time"

(b) "Comparing rating from different periods of time are meaningless whatever you do, and there is no chance in finding any formula that can compensate for possible rating distortions over time, since there are absolutely no grounds for such a comparison"

While I agree with (a) of course, (and even commented you about it a few times when you Directly compared Kasparov and Carlsen's TPRs ;) ), I disagree with (b).

I understood that you claim for (b), and that's what I was arguing with you about.

<one can talk all one wants and speculate that one gets more or less "skills" per rating point as compared to before, but i can't think of any method of making near-scientifical tests for whatever hypothesis one has. and i haven't seen any serious suggestion from anybody else, either. please feel free to suggest something that can be tested.>

Well surely you heard about the method that is mentioned almost each time this subject is discussed:

To analyze games of players from the different periods, then if you can rank the true chess strength of the selected players based on those selected games, you could say that rating R1 in yr Y1 matches rating R2 in yr Y2, and maybe come up with some general formula, if you can find a pattern there.

As for my suggestion for such analysis:

I agree that using a standalone engine for such analysis is a complete joke, I'd trust the general evaluation of an 2000 human player over Rybka's evaluation.

What I would trust is a team of 5 players from the top 10 (preferably from the top 5..), equipped with 10 of the best engines with strong hardware. Then take a selection of games where you know: The timestamps of the moves (or at least when there was time trouble), and when exactly was a novelty in each game (if we have the moves time stamps, we can also know whether the opponent had home preparation for the novelty or not).

The top players will rate the players level of play in the games considering: Their level preparation, ability to handle novelties OTB, time management, level of play in time pressure, ability to take advantage of opponents time pressure, ability to complicate when needed (also considering the opponent's weaknesses), pure blunders and strong moves (here the engines are very handy of course), endgame level of play and strategy, transitions from opening to middlegame and from middlegame to endgame, etc.

Apply voting mechanism on their conclusions.

Then, for further verification, we can compare their conclusions about rating distortions, with expected distortions resulting from changes in rating floors for example.

This will probably never be a real "scientific" method, but if the criteria are clear and the players are objective, then I would trust their evaluations.

I don't think you can fully scientifically prove that transitivity kept between sub pools divided by geographical and rating-level either. You don't have enough data there IMO, and surely not for the lower rated sub-pools. So you are applying some intuitive assumption and <speculations > in this case as well.

Jul-05-10  Etienne: An average performance for Carlsen...
Jul-09-10
Premium Chessgames Member
  Marmot PFL: But the average of his performances is steadily rising. Anand has said he will try to overtake Carlsen, but he has not much more chance than Armstrong of beating Contador.
Jul-08-12
Premium Chessgames Member
  SetNoEscapeOn: <Marmot PFL: But the average of his performances is steadily rising. Anand has said he will try to overtake Carlsen, but he has not much more chance than Armstrong of beating Contador>

And yet, it happened.

Jul-09-12
Premium Chessgames Member
  HeMateMe: What was this "King's Tournament"?
Jump to page #    (enter # from 1 to 30)
< Earlier Kibitzing  · PAGE 30 OF 30 ·  Later Kibitzing>
NOTE: You need to pick a username and password to post a reply. Getting your account takes less than a minute, totally anonymous, and 100% free--plus, it entitles you to features otherwise unavailable. Pick your username now and join the chessgames community!
If you already have an account, you should login now.
Please observe our posting guidelines:
  1. No obscene, racist, sexist, or profane language.
  2. No spamming, advertising, or duplicating posts.
  3. No personal attacks against other users.
  4. Nothing in violation of United States law.
Blow the Whistle See something that violates our rules? Blow the whistle and inform an administrator.


NOTE: Keep all discussion on the topic of this page. This forum is for this specific tournament and nothing else. If you want to discuss chess in general, or this site, you might try the Kibitzer's Café.
Messages posted by Chessgames members do not necessarily represent the views of Chessgames.com, its employees, or sponsors.
Spot an error? Please suggest your correction and help us eliminate database mistakes!


home | about | login | logout | F.A.Q. | your profile | preferences | Premium Membership | Kibitzer's Café | Biographer's Bistro | new kibitzing | chessforums | Tournament Index | Player Directory | World Chess Championships | Opening Explorer | Guess the Move | Game Collections | ChessBookie Game | Chessgames Challenge | Little ChessPartner | privacy notice | contact us
Copyright 2001-2014, Chessgames Services LLC
Web design & database development by 20/20 Technologies