Moscow Grand Prix R03: Plenty of Excitement!

by Alejandro Ramirez
5/14/2017 – The players came to round three with their fighting spirit reignited. Five decisive results and many hard fought games were seen in today's round in Mocsow. The most important for the standings, Ding Liren beat his compatriot, Hou Yifan, to take the early lead. He is followed by three of today's winners: Svidler, Salem and Mamedyarov. Results, analysis and interviews here...

The time control in the GP tournaments is 100 minutes for the first 40 moves, 50 minutes for the next 20 moves and then 15 minutes for the rest of the game plus an additional 30 seconds per move starting from move one.

The Grand Prix returns to the Telegraph Building in central Moscow, which previously hosted the 2016 Candidates Tournament won by Sergey Karjakin of Russia.

The tournament, a nine round Swiss contest, is the second of four Grand Prix in 2017 and follow’s the Sharjah Grand Prix in February which was won by Alexander Grischuk, Maxime Vachier-Lagrave and Shakhriyar Mamedyarov in a three way tie.

The Moscow Grand Prix is sponsored by Kaspersky Lab, PhosAgro and EG Capital Partners.

Each round starts at 2PM (GMT +3).

Round 3 on 2017/05/13 at 14:00

Bo. Name FED Rtg Pts. Result Pts. Name FED Rtg
1 Hou Yifan CHN 2652 0 - 1 Ding Liren CHN 2773
2 Vachier-Lagrave Maxime FRA 2795 1 ½ - ½ 1 Gelfand Boris ISR 2724
3 Tomashevsky Evgeny RUS 2696 1 ½ - ½ 1 Nakamura Hikaru USA 2786
4 Giri Anish NED 2785 1 ½ - ½ 1 Vallejo Pons Francisco ESP 2710
5 Mamedyarov Shakhriyar AZE 2772 1 1 - 0 1 Adams Michael ENG 2747
6 Harikrishna P. IND 2750 1 0 - 1 1 Svidler Peter RUS 2755
7 Nepomniachtchi Ian RUS 2751 1 0 - 1 1 Salem A.R. Saleh UAE 2633
8 Radjabov Teimour AZE 2710 1 ½ - ½ 1 Grischuk Alexander RUS 2750
9 Inarkiev Ernesto RUS 2727 ½ 0 - 1 ½ Hammer Jon Ludvig NOR 2621

All photos by Max Avdeev

The players have woken up in Moscow! A series of decisive games have left clear winners, and clear losers, in Russia.

Action packed today in Moscow

We will start with the draws of the day, which are the minority of the games.

First up is MVL facing against Gelfand. The Frenchman decided to repeat the variation that Giri used against the Israeli legend in round one, but Gelfand played his move order correctly this time and MVL didn't get more than a symbolic advantage as Gelfand said. The game ended soon after in a draw.

The Tomashevsky-Nakamura game seemed rather interesting, but in fact they were simply repeating a correspondence came from 2014. Tomashevsky presumably tested Nakamura's knowledge, which was clearly up to par. The game ended in a perpetual check that has been verified as completely sound by computers before the game even started.

Giri overstepped his limits trying to push an isolated queen's pawn position and found himself down a pawn. In the following interview Vallejo explains his decision to offer a draw:

However considering that the position was as follows:

Black is up a pawn and without any risking of really blundering anything. Even with a minute left on the clock surely Vallejo regrets not pushing this position.

Radjabov and Grischuk played a theoretical draw that had been seen before a few times. Their thoughts on memorization of draws:

Now to the juicy victories:

Hammer's Caro-Kann gave him a big edge

In the bottom board Inarkiev's 1.e4 was outclassed by Hammer's Caro-Kann. Strategically, already from a few moves out of the opening, Black's grasp on the kingside gave him a strong advantage. Hammer pushed his position forward and simplified into an endgame in which Black's beautiful knights were obviously much superior to an awkward bishop on f6:

Those are some pretty knights!

Mamedyarov annihilated Adams with a crushing attack:

[Event "FIDE Moscow Grand Prix 2017"] [Site "Moscow"] [Date "2017.05.14"] [Round "3"] [White "Mamedyarov, Shakhriyar"] [Black "Adams, Michael"] [Result "1-0"] [ECO "E52"] [WhiteElo "2772"] [BlackElo "2747"] [Annotator "alera"] [SetUp "1"] [FEN "rrb3k1/4qppp/pnpb4/8/1P1Pn3/3B1N2/2Q2PPP/1RB2RK1 w - - 0 19"] [PlyCount "23"] [EventDate "2017.??.??"] {[#]} 19. Bxe4 {Something has already gone wrong here for Adams. White's pressure on the entire board is clearly strong, and Black's counterplay is coming a bit late.} h6 $6 20. Re1 Qc7 $6 (20... Qf8 {is sad but necessary.} 21. Bxc6 $16) 21. Bh7+ Kf8 22. Ne5 $1 {White threatens simply Rb3 with a powerful attack on the kingside.} Nd5 $6 23. Nxf7 {A simple but effective combination} Qxf7 (23... Kxf7 24. Bg6+ Kf8 25. Re8#) (23... Nxb4 24. Nxd6 {threatens mate on e8}) 24. Bg6 Bf5 25. Bxf5 Nxb4 26. Qe4 Nd5 (26... Re8 27. Be6 Qf6 28. Rb3 { and Black is againt helpless to the rook transfer}) 27. Be6 {The game is over, the attack is too strong} Qf6 28. Rxb8+ Rxb8 29. Qh7 g5 30. Qg8+ 1-0

Salem Saleh played a beautiful game today against Nepomniachtchi. He put the Russian under pressure by sacrificing a pawn to open up his dark squared bishop and the h-file, causing real problems to the opponent's king. It is not easy to play when your king is constantly in danger, and in the following position, Nepo cracked:

Salem played a great game today

[Event "FIDE Moscow Grand Prix 2017"] [Site "Moscow"] [Date "2017.05.14"] [Round "3"] [White "Nepomniachtchi, Ian"] [Black "Salem, A R Saleh"] [Result "0-1"] [ECO "A12"] [WhiteElo "2751"] [BlackElo "2633"] [SetUp "1"] [FEN "8/1p2r1k1/2p3n1/5Q2/1PPb1P1q/5R2/7N/5B1K b - - 0 45"] [PlyCount "21"] [EventDate "2017.??.??"] 45... Rf7 {Clearly it is Black with the initiative, but there is nothing clear yet.} 46. Qe6 $2 (46. Qe4 Rxf4 (46... Bb6 {trying to keep some pressure. White should hold, however.}) 47. Rxf4 Qxf4 {would be just a draw, but Salem could have tried}) 46... Ne5 $1 {Nice! This tempo is crucial as White has no good way of protecting his rook} 47. Rh3 (47. fxe5 Rxf3 {is just an easy win for black, up material with the attack}) 47... Qxf4 {White is getting mated, and he doesn't even have a check} 48. Qe8 Qe4+ 49. Nf3 Rxf3 50. Qh8+ Kf7 51. Rh7+ { White hopes for a perpetual, but Salem calculates accurately that his king escapes with lethal consequences for the opponent} Ke6 52. Qc8+ Kf6 53. Qf8+ Kg5 54. Qh6+ Kf5 55. Qf8+ Kg4 0-1

Nepo has had three decisive results, all of which favored black.
Not a good thing for him, though, as he has had two whites.

Testing Svidler's knowledge of the Grunfeld isn't always a good idea. Harikrishna played into a very theoretically sharp variation, but Svidler navigated it masterfully and obtained a decisive advantage early on. Despite making his life a bit harder than it needed to be, his win was never in question.

Svidler showed, yet again, his magnificent handling of the Grunfeld

The game of the day as far as standings are concerned was certainly the Chinese duel. However, is was rather one sided. Hou Yifan's understanding of the opening was not the best, and she saw herself under pressure since then

Board one's duel was clearly in favor of Ding Liren the entire game

[Event "FIDE Moscow Grand Prix 2017"] [Site "Moscow"] [Date "2017.05.14"] [Round "3"] [White "Hou, Yifan"] [Black "Ding, Liren"] [Result "0-1"] [ECO "C54"] [WhiteElo "2652"] [BlackElo "2773"] [Annotator "Ramirez Alvarez,Alejandro"] [PlyCount "102"] [EventDate "2017.??.??"] 1. e4 e5 2. Nf3 Nc6 3. Bc4 Nf6 4. d3 Bc5 5. O-O O-O 6. a4 d6 7. c3 a5 8. Bg5 h6 9. Bh4 g5 10. Bg3 Kg7 {This idea of Kg7 is rather tricky. One of the main points is that the knight on f6 will be defended in certain variations.} 11. Re1 g4 12. Bh4 $5 {White sacrifices a piece, but Black is under no obligation to take it.} Ne7 $1 (12... gxf3 13. Qxf3 Be6 {and the pressure on f6 is annoying and will last quite some time. White can usually bail out with Qg3-f3 if he wants to, to force Kh7-g7.}) 13. Bxf6+ {Dragging the king to the center looks logical, but perhaps it is not best} (13. d4 Ng6 (13... Bb6 14. dxe5 { obviously doesn't work now}) 14. Nxe5 $1 (14. dxc5 gxf3 $17) 14... Nxh4 15. Nxf7 {is just a huge mess}) 13... Kxf6 14. d4 $6 Bb6 15. Nh4 Kg7 {The weird part of this position is that Black is simply better. The pair of bishops, the pressure on d4, the awkward knight on h4. It's just difficult for White to hold everything in an appropiate way. Hou Yifan decides it is time to shed some material to gain compensation.} 16. Na3 exd4 17. cxd4 Nc6 18. Nf5+ $5 Bxf5 19. exf5 h5 20. Nc2 Qf6 {The double attack was obvious, but White is hoping to create counterpressure.} 21. Re4 Qxf5 22. Bd3 Qg5 {White is fighting back, trying to create an initiative with active pieces to compensate for the pawn.} 23. g3 f5 24. Rf4 Rae8 25. h4 gxh3 26. Qf3 d5 27. Rd1 $2 (27. Rh4 $1 Kh6 $5 { and the game is still far from over}) 27... Re4 $1 {A typical but obvious sacrifice. White must accept the exchange sac but the resulting endgame is very unpleasant.} 28. Bxe4 fxe4 29. Qe3 Rxf4 30. Qxf4 Qxf4 31. gxf4 Ne7 $6 ( 31... Kf6 $1 32. Kh2 Nb4 $1 33. Nxb4 axb4 {with Kf5 coming and that is simply too many pawns.}) 32. Kh2 Ng6 33. f5 Nf4 34. f3 $6 {this gives Black another passed pawn} (34. b4 axb4 35. Nxb4 c6 36. Nc2 {at least attempts to bring the rook back into the game}) 34... c6 35. fxe4 dxe4 36. Re1 Bc7 37. Rg1+ Kf7 38. Rf1 Kf6 {Now it is really over. Black's pieces dominate and there are too many passed pawns for White to handle.} 39. Kg3 Kxf5 40. Ne3+ Kg5 41. Nc4 h4+ 42. Kf2 Nd3+ 43. Ke2 Bf4 44. Nxa5 h2 45. Nxb7 Nc1+ 46. Kf2 e3+ 47. Kg2 e2 48. Re1 Bd2 49. Rh1 Nb3 50. Kxh2 e1=Q 51. Rxe1 Bxe1 0-1

Ding Liren claims the lead in Moscow, half a point ahead of Svidler, Salem and Mamedyarov.

Round three games

Standings after three rounds

Click here for the full table

Round Four Pairings

Bo. Name FED Rtg Pts. Result Pts. Name FED Rtg
1 Ding Liren CHN 2773   2 Svidler Peter RUS 2755
2 Salem A.R. Saleh UAE 2633 2   2 Mamedyarov Shakhriyar AZE 2772
3 Grischuk Alexander RUS 2750   Vachier-Lagrave Maxime FRA 2795
4 Nakamura Hikaru USA 2786   Hou Yifan CHN 2652
5 Hammer Jon Ludvig NOR 2621   Giri Anish NED 2785
6 Gelfand Boris ISR 2724   Tomashevsky Evgeny RUS 2696
7 Vallejo Pons Francisco ESP 2710   Radjabov Teimour AZE 2710
8 Adams Michael ENG 2747 1   1 Nepomniachtchi Ian RUS 2751
9 Inarkiev Ernesto RUS 2727 ½   1 Harikrishna P. IND 2750

Live Commentary

AGON is offering exclusive pay-per-view video of the games and live commentary. It comes in three packages: a one-time $10 fee just for Moscow GP, a full package of all the events in the World Championship cycle for $30, and a $250 package, which is the same as the $30 Base but comes with signed posters from each event.

For more information, see the widget on the main page.


The games are being broadcast live on the official web site and on the chess server If you are not a member you can download a free Playchess client there and get immediate access. You can also use ChessBase 14 or any of our Fritz compatible chess programs.

Grandmaster Alejandro Ramirez has been playing tournament chess since 1998. His accomplishments include qualifying for the 2004 and 2013 World Cups as well as playing for Costa Rica in the 2002, 2004 and 2008 Olympiads. He currently has a rating of 2583 and is author of a number of popular and critically acclaimed ChessBase-DVDs.
Feedback and mail to our news service Please use this account if you want to contribute to or comment on our news page service


Rules for reader comments


Not registered yet? Register

turok turok 5/14/2017 09:54
first off I know this will piss many off but I am so tired of people making comments about how these GMs are not going for wins etc. Give me a break. They have worked hard and I mean very hard to get their rating where it is and it is NOT their job to please any of us in the cheap seats. Draws are part of the game. If they decide a draw is needed then so be it. Their style is their style. Now I understand in the past with Fischer when they conspired but if 2 players draw early that is their business unless there is an issue with giving points away and that is the person in charge to do. But to tell a player who has his rating for a reason to take chances because a commentator or chess fan thinks they should is ridiculous. it is hard getting points and even draws vs lowers hurt. If you want something to complain about complain about these sort of tournaments that these so-called Super GMs just swap points. That is the joke in all of this because of you had more tournaments with more GM ratings then those people can draw or beat these inflated super GMs and bring back these rodiculous over inflated ratings to the group.
Leonilo Leonilo 5/15/2017 12:16
@turok, you are not alone, I fully agree with you!
Atherton Atherton 5/15/2017 12:17
Valid point re the super GM tourneys turok
rubinsteinak rubinsteinak 5/15/2017 05:50
So...where is Carlsen again?
Mark S Mark S 5/15/2017 07:29
@rubinsteinak Why are you asking GM So on the whereabouts of World Champ GM Carlsen, when the other is in US and the world champ at Norway.
drcloak drcloak 5/15/2017 08:41

genem genem 5/15/2017 09:23
@turok . I agree with you that it is silly to blame the players for taking early draws. BUT I feel the Tournament Organizers have erred in not disallowing draw offers before the first time control after move pair 40. With of course, the necessary corollary rule that a player loses if his move creates a third occurrence of a position before move pair 40 is completed. The Vallejo Pons & Giri game should have continued, for the sake of the fans, for the more interesting game for everyone to replay and analyze later, for the decisive value of the tournament, and for the spirit of fighting chess.
lajosarpad lajosarpad 5/15/2017 10:38
@Turok I agree.
naisortep naisortep 5/15/2017 11:08
If a player is paying an entrance fee than he has a right to take a draw whenever he wants. But if you are receiving money from the sponsor than you have a responsibility to play a 'real' game. If it is a fighting game that happens to end in a draw that is fine. But if it is a prearranged or short draw then sponsors should take note. Invite players who will give the sponsor their money's worth.
naisortep naisortep 5/15/2017 11:16
Fans can root for whomever they choose. If they don't like a player because he takes too many short draws that is fine. It is like a basketball fan not liking a player who doesn't want to take the game winning shot. Maybe the player doesn't care what the fans want but the sponsors should be concerned.
blueflare blueflare 5/15/2017 12:11
mark, your response must be a joke i like it :)
KevinC KevinC 5/15/2017 02:05
@turok, I guess I will be the first to disagree somewhat. First, they may not answer to us, but the companies, which sponsor events are sponsoring it to get seen, and get good press. When GMs do this, they disappoint the sponsors, who in this case are trying (LOL) to charge for the privilege of watching the (CRAP) games, so indirectly, they do answer to us.

Second, to me, there is a difference between a draw. which was contested, and a 15-move GM draw. I have no problem with the former at all, but the latter is just one of the reasons chessplayers will never get paid like tennis players. If you really don't believe that it is their job to entertain us, then you have never thought about where the money that pays them REALLY comes from. This is the same for every sport.
geraldsky geraldsky 5/15/2017 04:05
if you want to see many draws...look for Anish Giri
koko48 koko48 5/15/2017 05:08
"first off I know this will piss many off but I am so tired of people making comments about how these GMs are not going for wins etc. Give me a break. They have worked hard and I mean very hard to get their rating where it is and it is NOT their job to please any of us in the cheap seats"

No other sport in the world has as many unplayed, non-games as chess. None.

The players owe it to us in the 'cheap seats' to play a real game, because the fans support the sport. Prearranged draws are a stain to the integrity of the sport, as well as the fan appeal.

The 3-1-0 scoring system was introduced to football in the 1980's with real and positive results. The few chess tournaments that used that scoring system ('Bilbao System') also saw a noticeable increase in the number of decisive games and fighting draws, with almost no non-games played.

Incidentally, the 3-1-0 scoring system was introduced in football to not only give more incentive for a win, but to prevent two teams "colluding" for a draw when their tournament placing gave them incentive to do so. It was successful at both. In football, mutually playing for a draw is called what it is.... "Collusion". In chess it's not considered cheating at all. It's just "normal", and "the way chess has always been played"

This is how far we've fallen backward in the chess world, that we accept this garbage as valid. In any other sport it's called what it is (and what Fischer called it): Cheating and collusion

So this round we got a lot of decisive games. Hooray. Too bad that's an infrequent anomaly in elite chess, and something that needs to be celebrated. THAT's something that should be the norm.

And part of the problem is "they worked hard to get their rating where it is". Yes, and now they feel the need to protect their ratings at all costs, by not playing chess games or taking risks - so they can keep getting invitations.
fons fons 5/15/2017 10:16
Chess will never be as popular as football, no matter what you do. Deal with it.

There are always a multitude of tournaments going on at the same time, what's the problem?
koko48 koko48 5/16/2017 01:34
@fons I mentioned this before in the comments to the Forbes article, but in 1975 the winner's share alone of the $5 million Fischer-Karpov WC purse was probably more than the highest paid football player, the highest paid baseball player ($240K per year), the highest paid American football player, the highest paid NBA player, and the highest paid tennis and golf players probably made that year - COMBINED

Will that ever happen again? In my opinion, no. But that proved that there can be a wide audience for chess under the right circumstances

Chess is the most popular board game in the world. Millions of people around the world know the rules and play at least casually. That is a huge potential fan base that has never been adequately tapped (except during the Fischer boom). And I believe prearranged draws (which at root, are caused and motivated by the traditional scoring system) are the main culprit
lajosarpad lajosarpad 5/16/2017 12:04

"The 3-1-0 scoring system was introduced to football in the 1980's with real and positive results. The few chess tournaments that used that scoring system ('Bilbao System') also saw a noticeable increase in the number of decisive games and fighting draws, with almost no non-games played. "

Can you found your statement on preferably a proof, but at least on statistical arguments, showing us how much did the 3-1-0 scoring system increase the number of decisive games and fighting draws? There was a debate at Chessbase where you argued based on the very same opinion, but your arguments were not convincing for me:
koko48 koko48 5/16/2017 03:47
@lajosarpad I'll tell you what I told the last person who questioned me on this. LOOK AT THE GAMES

The proof is in the pudding. I can't look at the games for you and I can't press 'google' for you.

In that thread you linked, I actually provided the questioner with the names of specific tournaments that used the Bilbao system. All he had to do (and all you have to do) is go on, search those tournaments, and look at the games.

The last questioner didn't bother to do that, even after he asked for proof. I hope you will not repeat that folly, and at least take the time to look at the games before arguing further. There is no 'statistical argument' that can make the case for football scoring as clearly and eloquently as the games themselves
drcloak drcloak 5/16/2017 11:51

I'd like some proof that you actually have any shred of common sense. All you know how to do is ask people for proof, so I've concluded you're a troll.
Petrarlsen Petrarlsen 5/17/2017 01:01
"(...) so I've concluded you're a troll" (drcloak). Says the most well-known troll of the ChessBase website... Eligible for the "comment of the year" award ???
Petrarlsen Petrarlsen 5/17/2017 07:19
I will continue the discussion on the "3 - 1 scoring system" on this page : (as I think that it is more logical to continue this debate on a more recent page about this same tournament).
lajosarpad lajosarpad 5/17/2017 02:09

I have been following high level chess for many years now and while I cannot claim that I have looked at all the high-level games played, I have seen a lot of games played in tournaments using the Bilbao scoring system. Even though this is completely irrelevant in our discussion, I am giving you this information to avoid any confusions about the fact that I have already looked at the games. What IS relevant in this case is the factual data. "Look at the games" is relying on subjective feelings, yet factually we can measure the average length of the games and compare it to the average length of games used at tournaments with more than less the same players using the traditional scoring system and compare the two. We can also compare the ratio of decisive games played using the Bilbao scoring system against those not using it. And we can compare individuals' game length and decisive game ratios under and not under the Bilbao scoring system, to get an idea of how is a given individual performing differently under Bilbao rules than under classical rules. These methods will provide us factual data and we need the largest possible samples. A given tournament will not prove your point. I can give you a tournament played in the 19th century with a lot of decisive games and ask you to show me a tournament played using the Bilbao system where the ratio of decisive games is at least as much. I am not doing that, as it would be incorrect. Just as much as getting a given tournament with a more than average decisive game amount using the Bilbao system and using it as "proof". I have asked you whether you have some factual statistics to support your hypothesis or is it just an opinion. Because you need to prove your statement or at least found it with strong arguments if you want me or anybody else to accept it. And no, sending us to "look at the games" is not a proof, not a strong argument, not even an argument. The first thing one who wants to prove something about this has to do is to form an educated guess, that is do the math.


If asking for a proof when a scientific statement is given is trolling, then most of the scientists are trolls and I am not in a bad company.
fons fons 5/18/2017 01:56
There are some statistics in this article:
but they only compare one tournament (Bilbao 2007) with others.
Petrarlsen Petrarlsen 5/18/2017 05:38
@ fons : In principle, the statistics from this article seemed very interesting.

But, unfortunately (and this is not at all your fault), these statistics seem in fact to be incredibly flawed !

Why ? Because Bilbao 2007 (unless I am much mistaken...) was 1) a rapid tournament (25 mn. + 10 s.), and 2) a blindfold tournament ! Yes !!!! (Source :

I simply cannot understand how the first author of this article (Josu Fernández) can compare six "classical games" tournament with a "rapid and blindfold" tournament ! This is really completely unbelievable !! It really amounts to comparing apples with oranges !...

@ lajosarpad : What is your opinion about this ?
lajosarpad lajosarpad 5/18/2017 10:30

The article's idea to compare the results in the two scoring systems is good. The problem is that this is the only good thing about this comparison. The sample does not contain enough data to make it relevant. I cannot escape the thought that the writer of the article was incredibly biased and/or had a pre-concept before writing the article and tried to get his concept as a result. Let's assume it is my hallucination only. Anyway, when one creates statistics, the person needs to be unbiased and to want to find out the truth, whatever it is. We could pick a classical tournament with many decisive results (like and compare it o the most drawish Bilbao tournaments. The result will be that the Bilbao system supports draws and since we have all seen that such an invalid result can be reached with less-than-adequate research method, we need to conclude that the size of the sample must be very large and the researched group should differ only in the researched attribute set from the control group. If we take 1000 White professional boxers and make them fight with 1000 Black scientists we will conclude that Whites are better boxers, which might be true or false, yet, it is a conclusion reached by fallacious means.

Bad method in creating statistics could lead to various invalid hypothesis. Take a look at this tournament and "look at the games" :)
With bad methods one can "prove" based on the London 1851 tournament that short matches between the players will lead to hard fought games ("look at the (Immortal) game"), however, a statistics created with good methods will tell us otherwise. Statistics even done on a large sample and with the right method can not be used as a proof. Take the prime numbers, for example. If one takes a large sample of primes, not knowing that 2 is a prime as well, one could conclude that all prime numbers are odd. Yet it is untrue, as 2 is pair and prime. Since statistics work with samples, it is prone to erroneous results even when done rightly. Even then we should not accept these results as proven facts, as they are just strongly founded hypothesis. The approach of getting a statistic done with an unsatisfying method to create the illusion that it is a strongly founded hypothesis and then stating that weakly founded hypothesis to be true is the method I have encountered from the proponents of the Bilbao system so far. I am not stating that they are wrong when stating that the Bilbao system so far resulted in more fighting chess. I am merely stating that the arguments and statistics seen by me so far did not live up to my standards.
Petrarlsen Petrarlsen 5/18/2017 11:32
@ lajosarpad :

Quite interesting developments !

And we obviously agree on our conclusions on these statistics...

I quite like : "If we take 1000 White professional boxers and make them fight with 1000 Black scientists we will conclude that Whites are better boxers, which might be true or false, yet, it is a conclusion reached by fallacious means." ; it shows very well the absurdity of the thing...
fons fons 5/19/2017 04:25
@ Petrarlsen >> "I simply cannot understand how the first author of this article (Josu Fernández) can compare six "classical games" tournament with a "rapid and blindfold" tournament !"

Agreed. I would be interested to see a proper comparison. The article came up in a quick search and I did not notice it was just rapid & blindfold, I just saw the tables with the statistics.

It's hard to find a good comparison, though I did not spend much time on it.
Petrarlsen Petrarlsen 5/19/2017 04:50
@ fons :

"I would be interested to see a proper comparison."

Me too. Some persons think that I am completely opposed to the "3 - 1 scoring system", but this is not the case (even if I think that I prefer the classical system - this because I think that the so-called draw problem is very much exaggerated, and thus, that there is no desperate need to find an urgent solution to it). For me, the main problem is that the usefulness of this "3 - 1" system doesn't seems to have been satisfyingly proven for the moment. In my opinion, it stands to reason that, if this system doesn't give the expected result, it has no interest whatsoever. (Why would we implement such an important change, if its usefulness isn't proved ?)

The problem, too, for a comparison between "3 - 1" tournaments and normal tournaments is that a big gap, in terms of level, between the top-players of a tournaments and the bottom-players, always tend to decrease the draw level. So it is a little complicated to compare different tournaments (a tournament with, for example, all participants between 2750 and 2800, as the last Candidates Tournament, will nearly always have a much worse draw rate that a tournament with 2800+ players and 2600 to 2650 players).

I think that the best idea would be to compare tournaments with a very even level distribution ; otherwise, I think it would be very difficult to have a really reliable result. But the problem is that, if I remember well, nearly all of the "3 - 1" tournaments, to this day, had a wide variety of levels, for the participants...
lajosarpad lajosarpad 5/19/2017 11:15

we could compare tournaments with uneven distribution as well, but we must make sure that the researched group and the control group should have an uneven distribution in the same way.
Petrarlsen Petrarlsen 5/19/2017 11:28
@ lajosarpad :

"we could compare tournaments with uneven distribution as well, but we must make sure that the researched group and the control group should have an uneven distribution in the same way."

Yes obviously. But as several factors can intervene, I think it would be quite complicated to be sure (or nearly sure) that the two tournaments would be really comparable (for example, to take the gap between the top player and the bottom player would be quite insufficient : if there would be 1 2600 to 2650 player and 11 2750 to 2800 players, it would not at all be the same as if there would be 6 2600 to 2650 players, and 6 2750 to 2800 players, and I think it would not be easy at all not to leave out any factor).
lajosarpad lajosarpad 5/20/2017 11:14

Tournaments involving the same players, or the similar average ÉLŐ (but if the average is similar, we need to have similar maximum rating gap as well) could be compared. However, since one will never put his/her foot into the same river twice, I think it is difficult to compare tournaments effectively, but we would rather form clusters of players, having the size of two and do the comparison.

Method: Repeat the following a lot in order to achieve a significant statistic:

1. Take two players and take the classical games they have played against each-other under the classical point system and the Bilbao system.
2. Compare the games, preferably taking into account several factors (result, length, used time by the players) and use their weighted sum to achieve a numeric result.

When we have a lot of such comparisons, we should aggregate them and see how players in general play against each-other under the Bilbao system compared to the classical system. While I am not willing to do this work, as this is not my researching field, if I were to do this research, I would start my planning using this method and maybe do some slight changes to improve it.
Petrarlsen Petrarlsen 5/20/2017 01:06
@ lajosarpad :

- "Tournaments involving the same players, or the similar average ÉLŐ (but if the average is similar, we need to have similar maximum rating gap as well) could be compared."

I think this is a still a little more complicated :

For example, we can take two 12 players tournaments with a 2700 average and a 200 points gap :

1) 1 2600 player ; 1 2800 player ; 10 2680 to 2720 players (for an average of 2700).

2) 6 players between 2600 and 2620 (average 2610) and 6 players between 2780 and 2800 (average 2790).

Even if the average Elo and the rating gap are the same in these two tournaments, I don't think that the drawing percentage will be the same at all in these tournaments. This is why I think it is very complicated to use the results from tournaments with a big rating gap...

- Your following method is quite interesting, and certainly very reliable ; the problem with it is that it necessitate a large number of games played with the "3 - 1 system", and this isn't easy to obtain (nothing is perfect...).

- My idea would be, for a first step, to try to obtain rough statistical data using the tournaments that has been played with this system for a first evaluation of the effect of this system on the draw rate. But, obviously, even for an approximate result, it is necessary to use a rigorous method, otherwise, the results would be simply completely meaningless... And I don't know if it would be really possible to find an approach that would give a sufficiently reliable result with the data at our disposal... What do you think of this ?
lajosarpad lajosarpad 5/21/2017 07:59

Yes, we have different ratio of participants in different point groups in different cases. This is a phenomenon called "white noise" when we are doing sample analysis. White noise will always add some randomity to the result of the analysis and is a significant contributor to the total amount of deviation from reality. However, if there is a large sample and we have the luxury to compare very similar tournaments, then the amount of white noise can be reduced into acceptable bounds. You will always have white noise, even if the comparison is perfect in terms of rating. One of the players might be divorcing from his wife, changing the attitude in one of the tournaments. Or different players with the same rating could be playing the two tournaments. We cannot put our foot into the same river twice (as second time the river has slightly changed its essence compared to the first time). I was referring precisely the white noise we have when I stated that it is difficult to compare tournaments. Not impossible, but difficult. This is why I have proposed a different approach and I think the current amount of games played under the Bilbao system should be large-enough to give us an idea, it is not enough to make that idea very strong. So when I asked Koko48 whether he has some statistical results to support his claim I knew that in the case he has, it cannot be very relevant, but would be very interesting to see even a statistic with reduced relevancy. Unfortunately he did not have such a result and he continued cherry-picking his favorite tournaments.

"Your following method is quite interesting, and certainly very reliable ; the problem with it is that it necessitate a large number of games played with the "3 - 1 system", and this isn't easy to obtain (nothing is perfect...). "
My proposed method could not yield very relevant results in lack of enough games to form a large researched group in the moment, yet I think the method can yield some interesting results with reduced scientific impact for now and if there will be many games played under the Bilbao system, then it could yield relevant results. Of course, this method should not be the final word and it is advisable to search for flaws in it and do improvements, or even to come up with better methods, as this method was only proposed as part of a conversation and not as the result of a rigorous perequisite analysis for a research. However, my method and any method yielding statistics, using a pattern to represent a larger group should, regardless of the result form only the strong foundation of a hypothesis and should never be considered to be a proof, as that would be a mistaken form of mathematical induction (the step of proving that an implication iterates the whole researched group is omitted).

I agree that we cannot have a reliable result in this moment. If we use the second approach I have shown (with similar maximum rating gap and average ÉLŐ), this could be a valid approach as well, if:

1. There is a very very large amount of games both in the research and in the control group
2. The white noise is random (that is, the white noises will not add up as a tendency)

However, this approach is much weaker than my proposal, as there are more potential white noises and therefore we need much larger samples and a lot of perequisite analysis to make sure that the result, whatever it may be is reliable. As for the "comparison" we have seen, it was a statistic created an a small sample, deliberately adding a lot of white noise to it to "prove" a point. I am allergic to this kind of unscience regardless of the author, the purpose and the result.