Greengard's ChessNinja.com

New View

| Permalink | 75 comments

Well, all of my worldly possessions have moved 200 meters to the east and my wife, daughter, two cats, and computer are still in working order. The new place is a disaster area of boxes and close to 2000 books (around half of the chess variety, the ones you're helping me get rid of). I'm battered, bruised, bleeding and bexausted. But it's a great new space, baby has her own room at last, and the air conditioner is in. That last just led to the discovery that the electrical system in this building is a horror show; apparently the landlord isn't much on maintenance. But at least the 3rd-floor walk-up will have me slimmed down in no time, whew.

Meanwhile, during all this I had to push a deadline for a 3500-word World Affairs article I'm working on with Garry, so I'm still not going to be around here too much this week. Then the ICC Chess.FM live coverage of the NH Tournament in Amsterdam (old lions vs young tigers, with Russian champ Svidler and US champ Nakamura starring) starts on Thursday. Maybe by then we'll have found a few other important things around here, such as the remote controls, the bathmat, and my underwear. Sweaty. At least we found the coffee maker 'cos I'm gonna need it.

Speaking of substance abuse, where's FIDE with their piss cups when you need them in London? Either Nigel Short is on something or the Dutch players he was annihilating at the Staunton Memorial found a Dutch coffee house near the tournament hall. The 44-year-old former world championship challenger just finished the team event with a bang, beating van Wely with black to rack up an amazing 8/10 score. I didn't think I was going to be missing much with this event, but now I'm sad my moving madness precluded a closer perusal. Short was the only player on the UK squad with a plus score, but his massive +6 (sorry ladies, he's married) made his team a winner, 26.5-23.5. Smeets led the Dutch on +2 and Sokolov had +1. Werle was the goat, losing five games without a win.

Some veterans who make Short look like a sprog were dishing it out as well over in the accompanying round-robin event. An inspired Jan Timman scored +5 to take clear first despite a loss to someone even older, Philidor's old sparring partner Viktor Korchnoi. The legend, now spinning at 78 rpm, took third place behind Cherniaev, who is only 40.

Down Jermuk way, the FIDE Grand Prix event in the Armenian resort town has exploded into activity from what I can tell. Aronian fell off the lead with two losses while Leko has remained steady and Ivanchuk has surged to take a share of the lead with +3. Kasimjanov is right behind on +2 after 8 of 13 rounds have been played. Ivanchuk's wins are spectacular stuff, vintage Chucky. Leko barely held on against Jakovenko in yesterday's game, managing to hold a pawn-down endgame with a knight against a bishop, not an easy thing to do.

Kamsky looked close to a win against Cheparinov but maybe his passed b-pawn just isn't enough? What about 51..Nf2? Kamsky nabbed his first win of the event against tourney dog Inarkiev. Theory watch: Akopian and Bacrot rehashed 15 moves from their encounter in the last GP in the Accelerated Dragon. Akopian won that one, but here he was the first to vary with 16.Rad1 instead of 16.Nb3. White tried the same quick e5 push but here couldn't avoid rapid exchanges and a quick draw.


Mig, no elevator in your new place ?

Nope. Elevators are pretty rare in these old buildings in NYC. This three-unit building was built in 1920, the row house we were in before up the street was 1911. Built to last. The popular real-estate tout "a pre-war building!" actually refers to places built before WWI, not WWII!

I have friends in 5th-floor walk-ups so I'll try to think of them instead of whining about my two flights of stairs. Now that the move is over it's not bad. We did get the added excitement of hoisting the box-spring up through a window when it wouldn't fit up the narrow stairwell. 200ft of rope, two guys braced in the window, one guy guiding down below so it wouldn't hit the air conditioners sticking out, two other guys holding onto the belts of the guys in the window. A very NYC DIY moving experience.

Before I stop whining completely, the bottom of the air conditioner cut up my fingers badly enough that the fingerprint reader I use to log on to my laptop won't recognize me. Got a pained laugh out of that one last night. At least I can still type in my password, if gingerly.

Nice Alekhine-Botvinnik, Nottingham 1936, vibe to the Kamsky-Leko game today. You don't see Black forcing a perpetual check against a half-naked king before move 30 too often.


Just tell it one of your trademark "Mig" puns, it can't fail to recognize you then.

Order of importance : wife, daughter, two cats and computer. wife and daughter are understandable. two cats more important than computer ?

I know, I know, I'm a failure as a geek. I did think about that move order for a minute, however. (And the kitties might have to fend for themselves if I have to get the comp or them out in a fire.)

And I didn't mention my G1, aka "my preciousssss," on that list since it goes without saying we are never separated.

Hey Mig, u have a fingerprint reader to log in?? That's pretty cool! Must think about getting myself one.

It's built into my Lenovo laptop. Very handy (pun intended, cc!) because the software also lets you use a quick swipe for your various passwords online and such once they're all entered.

"Philidor's old sparring partner" - brilliant, Mig!

There were some great games. I found Timman-Korchnoi very pleasing. Very precise play with the bishops (to my amateur eye).
Korchnoi for world champ! He can rant very well, too, better than most of these younguns.


Chucky back in the top 10!!!

(see chess.liverating.org)


If i had to chose someone to challenge the winner of the Anand - Topa match , it would be him the guy.
I´d like to see Chucky playing that match, enjoying the games instead of being nervous , and winning the title .
That would be awesome , says the teenager living inside me.

Chucky + Zoloft = WCh

"Then the ICC Chess.FM live coverage of the NH Tournament in Amsterdam (old lions vs young tigers, with Russian champ Svidler and US champ Nakamura starring) starts on Thursday." Mig

Svidler, Heine Nielsen, Beliavsky, van Wely, & Ljubojevic vs

Nakamura, Caruana, Smeets, Stellwagen, & Yifan

Which team is gonna win? Please reply. I'd love to read your educated opinions. http://www.chessbase.com/newsdetail.asp?newsid=5687

Sorry to hijack your link , but at the end of that article there is a video with Korchnoi and Hort talking about losing.
Maybe i´m missing something important but anyone knows What in the name of God is Korchnoi´s problem with Caruana?
From my perspective it seems that Victor is just being a jerk with the young boy ,( if someone ever gives me the hand in the way Victor does he better use the other to write)and then playing the nice old grandpa in front of cameras.

Such a big fuss about this so-called Staunton Memorial! A handful of Elo 2500 masters, carefully selected to not be embarassed by an Armenian, Indian or Uzbek player along the process, played a deplorable tournament, and what do you know, the English authors-of-a-book-in-three-days won!!! Such a great performance, it's the first one after the 'great' Staunton defeated a French patzer 150 years ago! By the way, winning the laurels against an ailing Dutch team, don't they become automatically World Champions (the perennial obsession of Englishmen)?? Or are they waiting to win also against Seychelles chess grandmasters, to look a little bit more legitimate?
Mig, you disappointed me by regretting that you didn't have enough time to follow this 'event'...

Not sure what fuss you're referring to. Two mentions of the event and one discussing the site's pay-per-view model.

Short's a former WCh challenger and just had a phenomenal result by ANY standard. And name his last book.

Steady on Vijay.

"don't they become automatically World Champions (the perennial obsession of Englishmen)??"

Is there any road map that could allow Chucky to get a shot to the WCH in the actual cycle?

Staunton defeated Saint-Amant in 1843, so that was 166 years ago. Not "150 years ago!" as you mistakenly screeched.

I would be curious to know the same thing.

Nosher over 2700 once again. All's right with the world...

Not to mention the ELO average of Short's opponents was 2622, so they're not a "handful of Elo 2500 masters."

Agreed. Being top 30 in the world and over 2700 is a big deal. Just ask the American and Nakamura fans. I'm begging for someone to reply to my old vs. young post. Who's gonna win?

29 Short off 2705,6 +21,6 10 1 1965 id-card
30 Rublevsky off 2703,0 0 0 0 1974 id-card
31 Akopian off 2702,6 -9,4 9 1 1971 id-card
32 Wang Hao

Warning: Non Chess Post

There may be some irony here: Ga. animal lovers killed by pack of wild dogs (AP) http://news.yahoo.com/s/ap/20090819/ap_on_re_us/us_deadly_dog_attack

Emmm , sure: via World Cup , Grand Prix or as a rating nominee ,as we know he is one of the few capable to raise his ELO exponentially in a short period of time).
I don´t actually know about his chances, though , i lost track of the current challenger system when Topa qualifyed to play Anand ;).
I always resented that he was not present in San Luis .

I think Ivanchuk is screwed though if what frogbert posted on ratings was accurrate cause Ivanchuk came in low on the latest list which is the heavy comparision point.

Forget ratings -- he is 2703 on the current list and would probably have to be at least 2825-2830 in January to have a chance.

He still has a theoretical chance to qualify via the Grand Prix, but after one mediocre and one bad result, quite unlikely.

Remains the World Cup or organizer's wildcard.

"Forget ratings -- he is 2703 on the current list"

Wrong. Ivanchuk is 2749.4 and #10 on the live ratings.

If you could pick an 8 player Super GM (2700+ or top 32 in live ratings) tournament, would you pick? These are mine in order of preference: Anand, Carlsen, Gelfand, Nakamura, Shirov, Wang, Dominguez, Bacrot, Short. Let the flames begin. ;)

Svidler, Heine Nielsen, Beliavsky, van Wely, & Ljubojevic vs

Nakamura, Caruana, Smeets, Stellwagen, & Yifan

Since no one answered my who will win the vets vs the young question, I'll answer it. I'll take the young, because if the top 4 seeds tie, I think Yifan will outplay Ljubojevic.

Wrong. Ivanchuk is 2703 and #30 on the official July rating list.

Two players will qualify to the Candidates by rating, using the average of the July -09 and January -10 ratings.

For Ivanchuk to catch up with those he will have to catch up with (such as Kramnik) it will take a miracle, even though he has already gained back many points.

If *I* could stage an 8-player Super GM tournament, it would be Kasparov, Chucky, Karpov, Korchnoi, Fischer, Capablanca, Lasker, and Steinitz. Since I've obviously gone into the realm of fantasy, let's make it 12 and add Morphy, Alekhine, Tal, and Botvinnik. Double RR.

And to further the dream, each of the players come into the tournament at their absolute lifetime peak, free of any of their concentration-altering vices, including political concerns, and into each game well rested (no quick draws).

(Super GM Section B is an open Swiss where you can put all of your other wannabees, like Euwe, Spassky, Marshall, Carlsen, Bogoljubov, Fine, Anand, etc. Probably 128 easy.

****Note that I have not mentioned either Kramnik or Topalov in these groups. But they'd both be in the Swiss!)

The Karpov/Capablanca minimatch might be extremely boring, or extremely grinding -- and gratifying. I think Tal finishes above Botvinnik. I think Steinitz finishes near the bottom (even though he has the benefit of over 100 years of theory added to his strength). I think plus-4 probably wins, unless someone like Fischer or Kaspy goes on a tear; even Chucky or Karpov, at peak ability, could do so, as could others. Where would Morphy finish? Or Alekhine?
Who has most wins? losses? draws? decisive games?

Ah, crap, I just woke up. Sigh.


"[Rising Stars vs. Experience] Which team is gonna win?"

Be patient, wait for Europeans to wake up ,:) - I will give it a try: By strenghtening the experienced team, the organizers made an effort to get a more balanced match than last year's 33.5-16.5 for the youngsters. The old team even has a slight ELO advantage (average 2642 vs. 2637). I would still favor the young team, for two reasons:
- Some of them are probably underrated at present (Hou Yifan, Caruana, Nakamura?)
- They are probably more motivated, also because the young player with the best result gets the prestigious Amber invitation.
This "competition within the competition" is also interesting. Who will mak it, Nakamura or Caruana? I wouldn't automatically favor the higher-rated player.

Speaking about motivation, for the other team a lot might depend on "which Svidler" is playing: sometimes he draws most of his games (quickly), sometimes he wins the Russian Championship ,:)

Korchnoi is a living legend, he is (or at least was) a great player. He is also a colorful and at times difficult character - going far back in time, it also took two to create a tense atmosphere during the Karpov-Korchnoi matches. I will not further elaborate on the [Cold War] context ... .

Korchnoi likes to be treated with respect, I have seen it myself when he participated in some blitz tournaments in the Netherlands. Maybe he feels (subjectively or not) that Caruana doesn't give him that kind of respect?

Still, it is a pity that he is not playing this year, as he always provided quotable statements to organizers and fans ... . Maybe the organizers should have invited Short this year, so Nakamura faces some competition in this field? ,:)

CO, in your top 8, I would definitely replace Korchnoi by Tal and Steinitz by Alekhine. And you forgot Smyslov! Also I think not 128 in the Swiss but say 16:

Spassky, Petrosian, Euwe, Steinitz, Korchnoi, Bronstein, Keres, Anand, Topalov, Ivanchuk, Kramnik, Shirov, Pillsbury, Smyslov, 2 out of the triumvirate of Bugoljubow/Fine/Andersson

And how about the top three in the Swiss joining the main double RR? :-)

Automatic entrants to main DRR: Fischer, Kasparov, Tal, Alekhine, Karpov, Capablanca, Botvinnik, Morphy, Lasker + top 3 from Swiss

Now that's a tournament worthy of inventing a time machine to hold!

Anyone looking for some juicy old chess gossip should check out Korchnoi's autobiography.

Maybe old Rufus is still around with that tardis thingy. Pity I don't have an oral chess history exam to pass in order to save the future.

Be excellent to each other, and party on, dudes!


I said the live ratings.

The live ratings are not official. They have no validity, and don't influence anything. They are purely for information purposes.
Ivanchuck's rating, for all purposes, is presently 2703.

I think FIDE needs to crawl out of the mud and use live ratings for everything. In the USA, batting averages for baseball players are updated every day. It would be laughable if people used a .250 batting average from May if the guy's current batting average in August was .385. I'm sure other sports keep their statistics up to date also. However, chess is in the dark ages, maybe even the stone age.

Live ratings might still influence "private" tournament invitations - after all, they are predictors for the next official FIDE rating. Ivanchuk is a bit unlucky that the upcoming September 2009 list does not count for the candidates tournament. But even with his present live rating of 2749 he is still behind Carlsen, Kramnik, Aronian, Leko and Radjabov (at most two of the last three names can qualify via the Grand Prix).

And while Manu is right that Ivanchuk is capable of "raising his rating exponentially", he is just as capable of losing many rating points in one tournament. Between now and the January list, he will play another GP tournament, maybe the World Cup and probably(!?) 1-3 other rated events.

So, like it or not, odds are that Chucky will remain one of the strongest chess players that didn't even get close to becoming world champion - if we discount his final match against Ponomariov in one of the FIDE KO events.

"I said the live ratings."

I said "he is 2703 on the current list" and he is, so you were wrong in saying I was wrong.

The point was that the context was Ivanchuk's chances to qualify to the Candidates by rating. It helps if you are reading the posts you are replying to. As I said, the official July rating and the January 2010 rating (the average of these two) is used to select two players to go there.

Yes, and folks need to understand that the "live ratings" are like chess "crack" -- they are fun to look at but not much use for anything.

Chess performances need to be evaluated over a fairly long period of time. A "hot streak" of 3, 4 or even 10 games should not be the deal-maker (or deal-breaker) for invitations.

Invitations -- or qualifications to the world title cycle -- should be based on a large body of games (or at least a fairly large amount of time or opportunity to play).

That's why they use the official lists instead of this "up-to-the-minute" useless ticker.

Think about it. When you take a college course, do you get a daily grade? Would you want a daily grade? Would it make sense to grade everything you do -- even homework -- so that the entire notion of "practice" vs. "test" disappears? Imagine if I said you were A on day 1, A- day 2, A+ day 3, B- day 4, B+ day 5, A- day 6...and so on. Useless!

Same with daily or game-by-game ratings. Plus, it distorts performance to the point where the measurment system (rating) can influence the game result that it is measuring (as in "I dare not make that sacrifice or I might lose half a point in the daily ratings!") Shades of the uncertainty principle in chess!

When you measure ratings that way...you are opening the measurement system (which is a token economy) to gambling effects. I'm sure everyone has seen the effect of throwing "good points after bad" in instant-rating online games. That's a gambling effect, whether the tokens are dollars in a slot machine or chess rating points on an online server.

That's not a good outcome for chess...which is why I never liked it when they changed FIDE ratings from the 5 point to the 1 point intervals...or when they sped up the production of "official" rating lists.

The top players in the world *shouldn't* change very often -- at least not over the course of 6-12 months. That's the point of believing in "chess strength" as a variable driving "chess performance."

"Korchnoi is a living legend, he is (or at least was) a great player. He is also a colorful and at times difficult character - going far back in time, it also took two to create a tense atmosphere during the Karpov-Korchnoi matches. I will not further elaborate on the [Cold War] context ..."

Too bad. I like attacking players like Fischer and Kasparov, but the Cold War made chess a cultural event. Watching Spassky/Fischer and Karpov battle Korchnoi (greatest post WW II non champ) and Kasparov were the greatest most exciting World Championships of my lifetime. I've never spoken to someone who was a chess fan before 86 who disagrees. There was a national and political tension to the matches you just don't see anymore. Computers played no part, because they were too weak. 69-85 was The Golden Age.

Mig, do you have a twitter account?

"Yes, and folks need to understand that the "live ratings" are like chess "crack" -- they are fun to look at but not much use for anything. Chess performances need to be evaluated over a fairly long period of time."

The live ratings are calculated the exact same way the official list is. They're unofficial, because they're not published by FIDE. Prove that Hans Arild Runde calculations are not based on FIDE's rating system. Let's see your mathematical formulas and proofs. It's just more up to date. That's why respected chess websites like Chessbase, TWIC, and this blog use them. Everyone knows this.

"The Live Top List is an unofficial chess rating list that intends to cover all players with a current, live rating above 2700 in the FIDE rating system. The official FIDE list is published 4 times each year, while this list is updated and published as games and tournaments are being played, typically with daily updates during tournaments featuring top players."

¨ Prove that Hans Arild Runde calculations are not based on FIDE's rating system. ¨

I believe you are mistaken , they should be the same calculations , with the only difference being the constant update.
There must be some software to do that , in fact FIDE should provide that software via download on their website ...
What a nice idea for a mac widget!

"I believe you are mistaken , they should be the same calculations , with the only difference being the constant update."

That doesn't even make sense. The same calculations being updated more often are still the same calculations.

Hence the ¨live¨ rating status of the thing.

Maybe Chessbase is doing a better job reporting on US Chess. They did report on the 2009 Denker Tournament of High School Champions. Those kids are great. It's hard for me to consider the WFM 2171 rated "Champ" the actual High School Champ when players like GM Hess and IM Robson are not playing though.

it was the weakest Denker seen in decades only 3 masters over 2200? None over 2250... only 5 experts? I mean a real pathetic showing of youngsters this year.

You make a good point. It's like...

Imagine someone listening to a play-by-play account of a basketball game on the radio: "Tim Bozo shoots and scores, and the Buffoons take a 87-86 lead with 1:46 to play. Hardy Shmardfarg is taking the ball up the court for the Wobblies. He passes to Poogie Crombockle. He shoots from the three point range..." Useless! Why would anyone want to listen to that when they can just look at the score in the newspaper in the morning!

And isn't it strange how people pay attention to baseball standings?! Right now, the Red Sox are seven games behind the Yankees. Who cares! At the end of the season I can check the final standings to see who wins the division.

Long post to follow, sorry for that - but I feel like responding to some nonsense regarding the live rating list ... :

Yep, to the best of my knowledge FIDE ratings and live ratings use the same algorithm for calculations. Actually there is no magic involved, the formula is available (for example in the Wikipedia article on ELO ratings). Anyone with time, motivation and a bit of computer knowledge (Excel should be enough) could do Hans Arild Runde's job, or double-check his calculations!? [Disclaimer: this is not meant to minimize his effort and experience]

The harder part may well be keeping track of all events where present and 'potential' 2700+ players are involved. He did "monitor" Nigel Short (FIDE 2684, live now 2705,6). Other names are, for example, Naiditsch (FIDE 2697 - but he will go further down after his Dortmund results) and Kasimdzhanov (FIDE 2672 - he will gain ELO after the Jermuk GP, probably not quite enough to cross 2700). Obviously, the live rating list can only cover top players [it would be rather impossible to keep track of all open tournaments and team competitions worldwide].

By contrast, FIDE is more 'passive' - processing only whatever rating report they receive. In principle, the FIDE and live ratings should converge whenever a new FIDE list appears, note that the FIDE lists are also just snapshots in time. I said 'in principle' because they won't if some tournaments do not submit rating reports in a timely fashion: Just two months ago, frogbert mentioned here that both MTel and Poikovsky were late in submitting their reports - and for a while it was unclear whether these events would be included in the July list.

So the live list emphasizes ongoing trends, and the current list (last updated August 19th) should be fairly similar to the upcoming September 2009 FIDE list. As far as top players are concerned: Four rounds of the Jermuk GP are still missing (as an official FIDE event, this will be rated in the upcoming list). Vachier-Lagrave has three more rounds in the French championship (which may or may not be rated for the next FIDE list).

@chesspride: At least in some situations players may have "ELO implications" in mind, even for a single game. In the last round of Dortmund, Kramnik may well have been aware that a victory against Naiditsch would give him 4,1 ELO points, while a draw would have cost him ~1 point. Minor differences, but they could become relevant for the rating spot into the candidates tournament.

"note that the FIDE lists are also just snapshots in time"

Exactly, FIDE's lists don't get more reliable or "correct" just because their updates are much slower and unpredictable with regards to what events they will include on a certain list. Rather the opposite since the one snapshot their lists are give much more importance to the situation at the exact moment when that list is calculated.

I disagree that FIDE lists are "unpredictable with regards to what events they will include" - this is (or at least should be) predictable, unless tournament organizers fail to submit their rating reports. For the rest I agree (with myself ,:) ). Normally it doesn't matter too much, but now for the candidates tournament two snapshots are unduly relevant compared to everything in between, including other FIDE snapshots. For example, Kramnik could also complain: he would be in better shape if Dortmund was played in June rather than July, or - conversely - if the rleevant FIDE snapshot was 1st August rather than 1st July.

What could be a better solution? Let's assume FIDE endorses the live rating list, should the tiebreaker be
1) the highest rating in a given period? This would favor 'erratic' players such as Ivanchuk, Shirov and Morozevich. Some people might like the idea, either on its own ground or because they dislike other players that are more stable and solid and/or play less tournaments.
2) the average of the highest and the lowest rating? Coincidentally(?), for Ivanchuk this would be roughly the same as his present live rating.
3) the average of all live ratings? Not necessarily the same as 2) above
4) the lowest rating in a given period!?

"this is (or at least should be) predictable"

It should be, but the rules are changed back and forth. I don't remember if it at the moment is 15 days between included events and list publication day, with the exception for events involved in World Championship cycles. This difference in itself hard to understand the reason behind. Previously there have been many instances where few have known what events will be included or not on the upcoming list, and FIDE have not always followed their own rules, but maybe they will improve.

"Yep, to the best of my knowledge FIDE ratings and live ratings use the same algorithm for calculations. Actually there is no magic involved, the formula is available (for example in the Wikipedia article on ELO ratings). Anyone with time, motivation and a bit of computer knowledge (Excel should be enough) could do Hans Arild Runde's job, or double-check his calculations!? [Disclaimer: this is not meant to minimize his effort and experience]"

I was a fan before, but I'm a bigger one now. It's difficult to reason with posters who make things up and scary how little they know.

"Exactly, FIDE's lists don't get more reliable or "correct" just because their updates are much slower and unpredictable with regards to what events they will include on a certain list. Rather the opposite since the one snapshot their lists are give much more importance to the situation at the exact moment when that list is calculated."

Exactly. I have seen situations where certain events are not counted for political reasons.

John come back to earth , like pointed by many (me first), the formula of the live ratings IS THE SAME that FIDE uses , it is you who were wrong.

I can still understand that an exception is made for "events involved in World Championship cycles" - actually I think it is 'official FIDE events' (generally the same, but it could also include Olympiads). FIDE can keep track of their own events (not too many after all), but conceivably not of every open tournament worldwide. But maybe they should also make an excption for round-robin tournaments category 18 or higher? [see below].

I found back the earlier discussion on this topic, it's in the "Ivanchuk Tired Enough to Win" thread posted June 21st. There frogbert wrote that the current deadline to submit rating reports is "seven days before the list date" [this may have changed again in the meantime ...]. This meant that the Bazna tournament finished one(!) day too late to be included in the July FIDE rating list - otherwise, Ivanchuk's current official rating would be 2720.

I see there was some mocking commentary of my criticism of the "live ratings list"...

...but folks rather missed my main point. The top players in the world don't change over a short period of time.

Jostling the ratings every game...is like brownian motion. Look -- he's on top. No, look, he's on top. No, now he's on top. Gibberish.

It is designed to put eyeballs on a list...not tease out truth about chess standings.

If you think the top 100 players in the world change position in the standings every game...I can't help you. Do they get stronger? Then weaker? What exactly are you trying to measure?

Ratings measure results -- that's what they measure directly. So if you are trying to quantify "he just lost a game" and "he just won a game" then the daily live list is for you.

If, however, you are trying to get at a variable like chess strength -- which we assume causally drives chess performance...

...then you need a long period of time -- a long stretch of results.

Otherwise, you are not capturing what is really interesting.

And yes it is the same rating system. But the rating system is not an end in itelf...it is (or should be) the means to answer the question "who is REALLY stronger" or "who would REALLY prevail in a match"

and such questions are not answered (or impacted) by small sets of games.

THAT is why ratings should not be viewed daily.

THAT is why ratings should not be viewed weekly

THAT is why even monthly blocks of data are suspect.

Not that the calculations are different...but that the effects you are trying to tease out are so small...that you can't see them that way.

For the fellow who mockingly said that we shouldn't care about baseball standings daily -- that only at the end of the season can we tell much...THAT IS RIGHT!

Teams that start out 20-14 can easily be 27-30 in a month and drop out of contention. Witness the Toronto Blue Jays this year -- leading the East at the start of the year and now having a fire sale on their players as they are sub-.500

All the daily standings do is sell newspapers (and I buy a paper just for that reason).

But you won't find out who is the best team in the NL by checking the daily results. Wait for the playoffs.

Just like you won't find out who is the better player by the live list -- wait for the quarterly rating lists...and even then...be prepared to track results for 2-3 years.


I believe you are right when you say that chess strenght cannot be meassured that often , but IMO having the live list serves so many other purposes that it would be weird not to have it.
Same thing with having a rating list for rapid chess, i know it shouldn´t be mixed with the classical rating but i still want to be able to check it out every now and then...
The live rating was a reality since the beggining of ELO , the only difference is that now you can look at them at any time.
IMO its like in any other sport ,sometimes you know the numbers and changes in the list without the help of anyone (soccer , rugby ) ,because the numbers are just easier to calculate , and sometimes you don´t (tennis , golf) ,that´s all.
You can choose wheter or not look at them but arguing against a number that is already in the air just don´t make sense.
However i agree with you that ¨official¨ ratings could appear even more sporadically , maybe only twice a year.

Not exactly. The quarterly lists can be viewed simply as samples, taken 4 times per year, of the live list (which you consider unreliable). Thus, they are no more statistically accurate than the live list at any randomly selected point in time. One might get a "less random" reading of strengths by applying an averaging filter to the live list, with the window length set to some reasonable duration. Something like a month or two, because that would allow and trends of legitimate improvement to be exposed while limiting fluctuating due to natural variations in performance.

Just wanted to point out that the quarterly lists don't include any such averaging, so they're obviously no better than the live list (which has been brought up here many times - "so and so got a lucky invite because he happened to be on a hot streak just prior to the official list coming out").

Of course, I should have mentioned that a similar effect to "averaging" or "smoothing" of ratings could be achieved to some extent by lowering the K-factor even further at higher ratings. I doubt this would fly unless there's a compelling reason to do it though.

Isn't the K-factor supposed to be set to whatever value maximizes its accuracy in predicting game outcomes based on rating? If it it's current value of 10 ( http://www.chessbase.com/newsdetail.asp?newsid=5376 ) is optimal or nearly so, there's no reason to change it.

I still remember those golden days last Autumn when No 1 of the live rating list changed daily. Every game had a huge stake, every move counted. Those were the days.

The point all you americana-style-flicker-number-adherents surprisingly fail to understand is, that there's an important difference between a long and a short term average. In statistics (as in science indeed) large bodies of sample data are essential in formulating results that can be trusted - indeed be of any use in any precise debate - and not just personal hunches. For that reason long term values are used within the realms of economy (quoting GDP per day would be _slightly_ imprecise - don't you think?), trade (same point with, say, the weekly banana export numbers from any Carribean country during the onslaught of a hurricane), climate (do you want to tell people western Scotland is a dry region because of one week without precipitation?), population, and so on ad infinite. Where and when one wants to express a general trend.

When talking accumulative averages - as we indeed are when talking rating - sports is a ill-chosen analogy, as usefulness and sober reporting since long has given in to public demand for daily excitement (and to an tragicomic degree in a US context): is it this or that player who is, as of today, the better batter (by 3rd decimal: oooooooh - big f...ing deal).
But it's only interesting quoting averages on a per-game-basis, when having a reasonable expectation that the numbers are expressing a clear and unmistakable long-term tendency.
And even then there's really not much connoisseurship/expertise involved in discussing (average) numbers when following sports. There are so many other perspectives on/qualities to a football, basketball or baseball game. Calculating and focusing on averages seems to be an outright waste of 'quality time' (a modern notion, the contents of which are probably unknown to most sports- and TV-addicts ;-). So within sports, THAT trend got it all wrong. Right?

Ivanchuk (for instance) is a most talented and exciting player, and I always love to play through his games. His ability to concoct spectacular motifs and beautiful positions (!) is rarely matched by peers. But as things are right now, I would want him to display stabilty over a longer period, before getting a shot at the WCh.
The WCh sould be at least one of the top3 players in the world - as opposed to a guy (or woman) that has had a one season hot streak (not that Ivanchuk couldn't come back rapidly and stay in the top for another few yrs!).

Think a little bit more about it - do you really want the ELO rating to express daily instead of periodic strength. Then take Shirov as one recent chilling example of such would-be rating-yo-yo'ism!

Summing up with ref @Thomas/hdghg: No, FIDE's lists are not snapshots in time - at least not in the normal sense of the word: a stocktaking on how things are right now/today. As a matter of fact they're meant to avoid that craze within the realm of chess - the 3/6 months sample period is a try to average out daily/weekly/monthly form. So that the list has more reliable information build into it.
As opposed to the live rating list, which is, by definition, much closer to expressing daily form. And of no use nor consequence, I'm afraid.
Indeed, when making _lists on the world's strongest players_, the problem to solve is finding the right balance between overly long and underly short sampling periods. Rating lists per decade? Nah! Rating lists per day/tournament? What's the use?

Short(let's see btw, how he performs 1 yr from now)-sightedness instead of a wider perspective. Epitomizing also, come to think of it, the creed of most politicians these days. A most dangerous and deplorable immaturity that's pushing our world towards the abyss. Sigh...

"No, FIDE's lists are not snapshots in time"

We will have to disagree about that one, I'm afraid.

"important difference between a long and a short term average."

Your post indicates that the FIDE list is a long term average and the live list is a short term average.

That's not the case. Both FIDE list and live list are weighted averages of all past results.

The calculation how the weighing is done, is the same. Both lists are snapshots of the same underlying "eternal" function. The FIDE list is a two-monthly snapshot of all players, while the live list is a daily snapshot of 2700+ players.

@Cynical Gripe
"The quarterly lists can be viewed simply as samples, taken 4 times per year, of the live list"

Not exactly. The live list uses the FIDE list as base for its calculations, not the other way round. When Aronian wins against Jakovenko, the live list uses their FIDE list ratings to calculate the change, not the live list ratings.

Your argumentation is fundamentally flawed: live ratings are NOT spot measurements, but updated long-term averages. Just two examples:

Climate: Let's say western Scotland had 10 consecutive years with 200 rainy days per year, followed by one year when (surprise, surprise) it didn't rain at all. The new "live precipitation index" will not be zero, but 181.8 rainy days per year (2000/11). In the language of climate scientists (I happen to be one) this comes down to the difference between weather [day by day] and climate [long-term averages].

Chess: If Ivanchuk (or anyone else) has a rating of 2750 and a TPR of 2900 (or 2600) in one tournament, his new live rating might become 2770 or 2730 [I won't bother to make exact calculations], not 2900 or 2600. This may be the difference between playing strength (or chess understanding) and current form.

The only difference between FIDE lists and live rating lists is that the former are calculated less often and thus depend more on Ivanchuk's form at exactly the moment when the list is made.

You're right. I overlooked that important point.

I see what you're saying, but the current system is still flawed.

Imagine two players with identical ratings of 2600 at the start of the rating period. Player 1 goes on a consistent hot streak and manages to reach 2700 by the end of the rating period. On the other hand, Player 2 starts with a couple bad performances, such that his live rating drops to 2550 at the halfway point, but turns things around with a couple good performances towards the end. He ends up with the same 2600 rating as he started with.

Beginning at the next rating period, Player 2 continues his good form and manages to reach 2650 before cooling off and dropping back down to 2600. Player 1 performs poorly throughout and ends up back down at 2600 again (or even slightly lower).

Overall, the two players have actually performed almost identically, not only in total wins and losses, but also in the length of their respective hot streaks. There's no arguing that Player 1 reached his peak of 2700 by showing legitimate, sustained improvement compared to player 2. It's just a fluke that the point in time when ratings were sampled coincided with his peak. Biased results despite built-in "long-term" averaging.

I also see your point, but I would call this a 'deficiency' of the current system rather than a 'flaw'. Flaw might suggest that there is something fundamentally wrong, and/or that things could be done in a different and better way.

I fail to see such a better way, still I can offer some thoughts: In statistics, every number has an associated uncertainty, so maybe this should be _explicitly_ added to anyone's ELO. Ivanchuk's rating may then be 2750+-50, Caruana's rating may be 2670 +50/-20. Formally these would be 68% confidence intervals for how their 'snapshot' ratings could develop in the nearby future (based on past evidence); Caruana's asymmetric distribution acknowledges that he showed some "legitimate, sustained improvement".
Of course things would get messy if we go all the way applying proper error propagation. Then in a tournament a player's performance rating may be 2750+-100 ... .

Anyway, from the above it should be clear that 2701 is not significantly higher than 2699, and 2753 is not significantly higher than 2750 (contrary to the belief of some organizers and chess fans!?). But going back to the rating spot for the candidates tournament: FIDE might end up concluding that numerous candidates have statistically indistinguishable ratings, so none or all of them should be invited.

Not a practical solution, so we must accept deficiencies of the current rating system. The only alternative I see is a radical one: no rating spots at all, instead invite the #3 of the Grand Prix and the losing finalist of the World Cup. But this would exclude some players that, for whatever reason, don't play those qualifying events.

Twitter Updates

    Follow me on Twitter



    About this Entry

    This page contains a single entry by Mig published on August 18, 2009 1:49 AM.

    Chess Junk Seeks Good Home was the previous entry in this blog.

    Tricks for Not So Old Dogs is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.