< Earlier Kibitzing · PAGE 12 OF 12 ·
Later Kibitzing> 
Jan2713   csmath: Wild game. Unusual KID. I am guessing that Giri was ready for a peaceful draw until he saw KID. I am amazed this ended up in a draw. 

Jan2713   Hopsa: <FSR:> What??? some patzer used my analysis before? 

Jan2713   csmath: One of the things to know when you analize KID is that engines do not evaluate it very well, and that is true even for the best engines. So be careful when making assessment based on engines in KID.
It is also extremely hard to find the best moves in KID over the board, even for the best players in the world. 

Jan2713
  lost in space: <<AylerKupp:> <lost in space> Carlsen's current rating is 2872.2 per http://www.2700chess.com. I found them to be remarkably fast in updating their ratings; I checked them yesterday just a few minutes after Carlsen's win over Nakamura and they had already updated their page. I don't remember what his rating was then but it was obviously higher before his draw against a lowerrated opponent.
And, BTW, Carlsen gained 11.2 rating points in the Tata Steel 2013 tournament. You can see how many points each tournament participant won or lost, and what their current world ranking is. Thanks <AK>, have checked it in the meantime. It so unreal to see a player running so far ahead of the file, 2900 seems to reachable for him this year, maybe already during thr candidates (if he keeps this form). 

Jan2713
  lost in space: Sorry, can not delete my last post any longer. Meant is "running so far ahead of the field..." 

Jan2713   solskytz: <Kinghunt: This is a King's Indian Defence  black is always objectively lost until suddenly black wins.> I loved that one too 

Jan2713   solskytz: The candidates is a 14round business.
Given a Kfactor of 5, what performance does Carlsen need there in order to blow through 2900? He will need to go up 28 points, or 2 points per game. If I counted right, his opponents are on the average, rated 2774.5 (as of current live ratings, assuming these will become official and won't change for these players until the candidates). Going up 2 points per game with a K factor of 5, means that you should perform at 70 percent against a field rated the same as yourself  which is the same thing as to say, outperforming your own rating by 149 points, according to the elo calculation table. Carlsen will need a performance of 2872 + 149 = 3023 in the candidates to break 2900. 3023 against a 2774.5 field, means playing at 248 points better than the field  meaning a result of 81%, meaning that 11.5 out of 14 (!!) will do.
That would be a record performance for him, but not by that much. It's rather unbelievable that he can pull it off, but hey, this is Carlsen we're talking about. If he manages to sweep the candidates that way, he'll start to remind people of Alekhine in San Remo (1930). He would then fully deserve a 2900 rating straight away, without further delays. 9.5 out of 14, which is very very credible, even a convenientlymargined result given his current form, is still a solid 2900+ performance (closer to 2910 actually), and will get him closer to the mark, and very justifiably so. In fact, 9 our 14, which is a stellar achievement in this field, will still justify his already humongous rating and not cost him any. Any result below that will cost him, I suspect, more than just elo points  I don't even want to think about that! I predict a 10point finish, and a subsequent WC match against Anand later this year. 

Jan2713   solskytz: 3021 performance needed, not 3023. Small arithmetic error, with no influence over the rest of my post. 

Jan2713   solskytz: By the way  in case Carlsen and somebody else finish on the same score (say, 9.5 or 9 out of 14), what is the tie break involved? Does anybody know? 

Jan2713
  AylerKupp: <<Lost in space> It so unreal to see a player running so far ahead of the field,> Actually he is nowhere close to being the all time furthest ahead of the field, Fischer holds that distinction. In 1971 at 2785 he was 125 rating points ahead of #2 Spassky at 2660, and in 1972 at 2780 he was 120 points ahead of #2 Spassky at 2660. Other notable gaps between #1 and #2 are:
1999: #1 Kasparov (2851) vs. #2 Anand (2769), an 82point gap 1973: #1 Fischer (2780) vs. #2 Karpov (2700), an 80point gap 1992: #1 Kasparov (2805) vs. #2 Karpov (2725), an 80point gap 1974: #1 Fischer (2780) vs. #2 Karpov (2705) a 75point gap 1990: #1 Kasparov (2800) vs. #2 Karpov (2725), a 75point gap 1989: #1 Kasparov (2800) vs. #2 Karpov (2730), a 70point gap So far in 2013 #1 Carlsen (2872) has "only" a 62point gap between himself and #2 Kramnik (2810). Still, a 62point gap between #1 and #2 is impressive, and more than 1000 points between him and my alltime high rating. :) 

Jan2713   fgh: <AylerKupp>: Here you go for the biggest gaps: http://www.chessbase.com/newsdetail... 

Jan2713
  AylerKupp: <solskytz> Actually, the lowest Kfactor that FIDE is currently using is K=10 for players rated > 2400. How does that change your calculations? 

Jan2713
  AylerKupp: <fgh> Thanks, I had forgotten that Sonas had done these calculations for players from other eras. But he was using his Chessmetrics rating system rather than the Elo rating system so his indicated 146 point rating gap in 1971 between Fischer and Spassky is not directly comparable to his 125 point rating gap using Elo ratings. Other than it was very impressive. BTW, this is the list I used for that 125 point rating gap: http://www.olimpbase.org/Elo/Elo197.... 

Jan2713   solskytz: <AylerKupp> well, yes  thanks for the correction, but that's actually what I meant... when I wrote '5' it's because with k=10, you get 5 points for beating somebody as strong as you... 

Jan2713
  morfishine: Hello <Ulhumbrus> Yes, your comment here reflects something I was wondering: <If Carlsen was playing White at move 31 he might have declined the bishop by 31 Ne2 or 31 Nh3 or (as <morphishine> pointed out ) by 31 Rb1 and then won a long ending with his bishop pair> Unfortunately, 31.Rb1 also draws, a pity 

Jan2713
  alexmagnus: Currently, Carlsen's peak gap to #2 on an official list is the 4th biggest (behind Fischer  125, Kasparov  82 and Karpov  65). But it surely won't be long till he surpasses Karpov  and quite probably Kasparov. Fischer's 125 are hard to touch. Same with gap to #10, where he is behind the "great trio" (Kasparov 175, Fischer 160, Karpov 130) and Kramnik (110) but will be ahead of everyone else on the coming list. Surpassing Kramnik and Karpov seems just a matter of time, while Kasparov and Fischer is quite a challenge. 

Jan2713   theagenbiteofinwit: What a blast from the past. The KID was one of Carlsen's first openings, but he stopped playing it after a while because he thought it didn't suit his playing style. 

Jan2713
  Eyal: It looks like 13...c5 (at least at the timing in which it was played by Carlsen) was a strategic mistake, and after Giri's very good reaction  14.dxc6 followed by 15.c5!  Black came under a lot of pressure. Giri himself said in the postgame interview (http://www.youtube.com/watch?v=xRvL...) that until he saw that sequence he wasn't sure how to handle the position on the most basic strategic level ("It was very complicated  I don't know, I want to close the Kside, or I want to close the Qside, or he wants to close the Kside, or he wants to close the Qside..."). 

Jan2713
  Eyal: Position after 27...Nd7:
click for larger viewduring the game, a lot of people who were using computer software were screaming for 28.Qa3, as they were after 28...Ne8 a move later. It's certainly the best move and probably wins, but as Daniel King notes in his video analysis of the game (http://www.youtube.com/watch?v=CM3P...) it's also terrifying to play such a move from a human point of view, because you have to be totally accurate in calculating the consequences of a move like Bxb3 for Black  get one little detail wrong and you might get checkmated. As it happens, 28.Qa3 Bxb3 29.axb3 Rxb3 doesn't work for Black after 30.Qa2 (pinning the rook) and now, for example, 30...Kh8 (unpinning it and threatening Qd6/e7 & Ra3) 31.Ne2! (preparing Rc3 in response). Keeping the queen on the 6th rank, as Giri did on moves 2829, had the practical justification of guaranteeing that he can exchange queens in case of Bxb3 and stay out of danger. 

Jan2713   csmath: <it's also terrifying to play such a move from a human point of view,> Yup. I would never play that. To put queen in the path toward king with all black pieces breathing down, protected by two pawns?
The variation might be winning but over the board to commit to such a decision? You gotta be cool as cucumber and no player short of perhaps Kramnik can play KID with cool head. Giri trying to place queen on "active" square seems a lot more reasonable over the board. Nowadays we see all by engine but in old times it could have been many months before somebody would come up with Qa3. 

Jan2813
  morfishine: Two noteworthy aspects of this game: (1) White never moves his King Knight nor his King Rook & (2) the Black DSB maneuver Bg7/f6/e7; Pure positional chess by Carlsen; Since Black's WSB forces the draw at the end, the question is does 19.Nxb6 axb6 20.Bxc8 improve for White? Something like this would activate White's King side while eliminating Black's WSB 

Jan2813
  perfidious: <Eyal> The idea 28.Qa3 readily suggests itself as a possibility, but I agree: it is far from easy to play in a practical game. Small wonder Giri bailed out instead. 

Jan2813
  perfidious: On another note, it was amusing to see that there were no similar games listedthe truth is rather different, as the position after 10....Bf6 is a wellknown line in the Saemisch KID by transposition which has scored well for White, who usually plays 11.Nge2, intending to meet 11....Bh4 with the strangelooking 12.Bg1, when his queenside attack will carry on despite long castling. Opening Explorer 

Jan2913
  AylerKupp: <solskytz> Oh, sorry about that, I often tend to take people too literally. But I must say that I don't understand the statement that Carlsen will need a performance of 2872 + 149 = 3021 in the candidates to break 2900. And given that Carlsen is currently rated 62 points ahead of Kramnik (the current #2 rated player) and every other tournament participant is rated less than that, it's clear that the field cannot be rated the same as Carlsen. Or maybe I'm interpreting things too literally again. However, we get similar results. This is how I calculated it, somewhat differently, and yet got a similar result. Please correct me if you see something I did wrong. 1. I calculated, like you did, that the average score of Carlsen's opponents to be 2774. Actually I got 2774.21; I'll display the rounded ratings but I used the unrounded ratings in the calculations, not that it makes much difference. 2. Per http://en.wikipedia.org/wiki/Elo_ra... a player's new rating is New Rating = Old Rating + K*(ASES) where: K = 10 for players rated above 2400
AS = Player's actual score in tournament/match
ES = Player's expected score in tournament/match
And
ES = 1 / 1 + 10^[ (Opponent Rating – Player Rating) / 400 ] So as an approximation (ES of course depends on the actual opponent's rating and if the ratings changes get calculated after every game): ES = 1 / 1 + 10^[ 2774 (average opponent rating) – 2872 (Carlsen's current rating) ] = 0.638. This ES is the same as the Elo P(Win). Then using the Elo P(Win) tables Carlsen would need to score at least 95 (0.643 or 64.3%) in order not to lose any rating points since AS = ES + (New Rating – Old Rating) / K. Using the same formula he would need to score 122 in order to reach 2903. Of course, this is an approximation. Each player is rated differently so Carlsen would get more rating points with a win against Kramnik (62 point rating difference) than with a win against Gelfand (132 point rating difference), and this could be significant if the ratings are recalculated after every round. So it's not only the number of wins, draws, and (gasp!) loses but who he achieves these results against. And in the final round, if Carlsen needed a draw to clinch the Candidates Championship but a win to break 2900, I'm pretty sure that he would not take unnecessary risks and choose the former. So you calculate that Carlsen would need to score 11.53.5 to reach 2900 and I calculate that he would need to score 122 to reach 2900. Pretty much the same thing given our approximations. But I personally don't think this would happen. 

Feb0113   solskytz: <Aylerkupp>
I would trust your method of calculation better than the one I used myself, as yours is based on the actual elo formula, while I was just using a good approximation. It's still nice that I got close... And I also don't suppose Carlsen will get 12/14  I think he'll make ten points and win with a margin of one point over Kramnik. 


< Earlier Kibitzing · PAGE 12 OF 12 ·
Later Kibitzing> 


