+GGT+ Deep Rybka 3 wins Lisitsin Gambit and takes the lead+

[rainhaus]

+GGT+ Deep Rybka 3 wins Lisitsin Gambit and takes the lead+

Gambit No 7
AO4 Lisitsin gambit

1.Nf3 f5 2.e4

Wikipedia:
Georgy Lisitsin (Russian: Георгий Лисицын; 11 October 1909 - 20 March 1972) was a Russian chess master.
He won thrice Leningrad City Chess Championship in 1933/34 (joint), 1939 and 1947 (joint). He participated many times in USSR Chess Championship. His best result was in 1933 when he shared 3rd (Mikhail Botvinnik won). He also played in international tournaments, finishing 15th at Moscow 1935 (Botvinnik and Salo Flohr won), and took 2nd at Helsinki 1946. Lisitsin was an author of several chess books. He was awarded the International Master title in 1950

Wikibooks:
Lisitsin Gambit
Black logically accepted White's gambit and is now attacking White's f3-knight. By far the best move to save it is 3. Ng5. On d4 or e5 it can be kicked by pawns, on h4 it does nothing (as long as Black is wary of a Qh5 check) but on g5 it forces Black to defend her e4-pawn.

Meanwhile, this gambit was modernized with the "Zwischenzug" (~in-between-move) 2.d3.. So played by 13-year-old Magnus Carlsen, Aeroflot Open 2004.
http://www.chessville.com/reviews/Secre ... prises.htm

Win%

Code: Select all

Games        :     42 (finished)

White Wins   :     14 (33.3 %)
Black Wins   :      9 (21.4 %)
Draws        :     19 (45.2 %)
Unfinished   :      0

White Perf.  : 56.0 %
Black Perf.  : 44.0 % 

A surprisingly high performance with this gambit, better than the human one! After 10 gambits I'll give a first performance ranking of the several gambits.

Lisitsin Gambit, Human Chess
Internet Database

Code: Select all

games       1-0     =-=     0-1    Score
------------------------------------------
1116        469     264     383    53.8%
-----------------------------------------------------------

Starting evaluation
After 2.e4:
0,00 to - 0,32 cp
Mean = - 0,15
Not good correlation with the excellent performance on the side of the gambit player.

Opening Table Lisitsin Gambit(SCID)

The gambit is accepted by all engines and they also follow the main line 3.Sg5

Opening Table. Moves 1-8 by all engines

Code: Select all

-------------------------------------------------------------------------------
  1.Nf3 f5: +14 =19 -9 (23.5/42: 56%)
-------------------------------------------------------------------------------
       2        3        4        5        6        7        8        9     
-------------------------------------------------------------------------------
 1  e4       Ng5      d3       Bd3      h4       h5       hg6      Rh8       4:
    fe4      Nf6[1]   ed3      g6[2]    d6       Bg7      hg6      Bh8[3]   63%

 2  ...      ...      ...      de4      Nc3[4]   Bd3[6]   Nf3      OO        7:
    ...      ...      Nc6      e6       Bc5[5]   h6       OO       d6[7]    57%

 3  ...      ...      ...      de4      Nc3[9]   Bb5[11]  OO       Nf3       8:
    ...      ...      e6       Nc6[8]   Bc5[10]  OO       h6[12]   d6[13]   38%

 4  ...      ...      d3       de4[14]  Qh5      Qh4      Nf3      Qh5       7:
    ...      d5       Qd6      h6       g6       Bg7      g5[15]   Kf8[16]  50%

 5  ...      ...      ...      Nc3      Nh3      de4      Qh5[18]  Qh4       4:
    ...      ...      ...      h6[17]   e5       de4      g6       Bf5[19]  63%

 6  ...      ...      ...      ...      ...      Nf4      de4[20]  Ng6       4:
    ...      ...      ...      ...      Nf6      Nc6      de4      Qd1[21]  50%

 7  ...      ...      ...      ...      Nb5      Qh5      Nc7      Na8       1:
    ...      ...      ...      ...      Qb6      g6       Kd7      Qa5[22]  100%

 8  ...      ...      ...      ...      ...      Nc7      Qh5      Nf7       7:
    ...      ...      ...      ...      Qc6      Qc7      Kd7      Nf6[23]  71%
-------------------------------------------------------------------------------

[1] 3...g6 4.c4 Nf6 5.Nc3 Nc6 6.g3 e6 7.d3 exd3 8.Bxd3 Bg7 9.Be3 OO 10.OO b6 0-1(53) Zappa Mexico II - FireBird 1.0 beta x64 mp, ? 2010.
3...Nc6 4.d3 Nf6 5.dxe4 e6 6.Bd3 Bc5 7.OO OO 8.Nc3 a6 9.Qe2 d6 10.Be3 Bxe3 =-=(90) Naum 4 MPx64 - Rybka 3, ? 2010.

[2] 5...Nc6 6.Nxh7 Ne5 7.Bf4 Nf7 8.Ng5 Nxg5 9.Bxg5 d6 10.OO c6 1-0(111) Naum 4 MPx64 - Deep Fritz 11, ? 2010.

[3] 10.Bxg6+ Kf8 1-0(57) Deep Shredder 12 x64 FrGui - Deep Fritz 11, ? 2010.

[4] 6.Bd3 Bc5 7.Nc3 b6 8.OO OO 9.Na4 Be7 10.f4 h6 1-0(51) Rybka 3 - FireBird 1.0 beta x64 mp, ? 2010; 7...a6 8.OO OO 9.Nf3 Ng4 10.Bg5 Qe8 1-0(94) Deep Fritz 11 - Stockfish 1.6 JA 64bit_dec09, ? 2010.

[5] 6...Bb4 7.Bd3 OO 8.OO Bxc3 9.bxc3 Qe7 10.Rb1 e5 0-1(75) Zappa Mexico II - Deep Shredder 12 x64 FrGui, ? 2010; 10.Qe1 e5 =-=(73) Stockfish 1.6 JA 64bit_dec09 - Deep Shredder 12 x64 FrGui, ? 2010.
6...h6 7.Nf3 Bc5 8.Bb5 OO 9.OO d6 10.Bxc6 bxc6 0-1(62) Zappa Mexico II - Rybka 3, ? 2010.

[6] 7.Bb5 h6 8.Nf3 d5 9.exd5 exd5 10.Qe2+ Qe7 =-=(100) Stockfish 1.6 JA 64bit_dec09 - Deep Fritz 11, ? 2010.

[7] 10.Na4 Bb6 1-0(100) FireBird 1.0 beta x64 mp - Rybka 3, ? 2010.

[8] 5...h6 6.Nh3 e5 7.Be2 Bc5 8.Bh5+ Kf8 9.Be3 Bxe3 10.fxe3 Kg8 0-1(43) FireBird 1.0 beta x64 mp - Deep Shredder 12 x64 FrGui, ? 2010; 6.Nf3 Nc6 7.Nc3 Bc5 8.Bd3 OO 9.OO Ng4 10.Qe2 Qf6 1-0(50) FireBird 1.0 beta x64 mp - Stockfish 1.6 JA 64bit_dec09, ? 2010.

[9] 6.Bd3 Bc5 7.Nc3 h6 8.Nf3 OO 9.OO d5 10.Qe2 a6 1-0(189) Rybka 3 - Deep Shredder 12 x64 FrGui, ? 2010.

[10] 6...Bb4 7.Bd3 OO 8.OO Bxc3 9.bxc3 Qe7 10.Qe2 e5 0-1(69) Deep Fritz 11 - Deep Shredder 12 x64 FrGui, ? 2010.

[11] 7.Na4 Bb4+ 8.c3 Be7 9.Bd3 Rb8 10.c4 OO 0-1(63) Deep Shredder 12 x64 FrGui - Rybka 3, ? 2010.
7.Bd3 Qe7 8.OO a6 9.Nf3 d6 10.Qe2 Bd7 =-=(45) Deep Fritz 11 - FireBird 1.0 beta x64 mp, ? 2010.

[12] 8...Qe8 9.Qe2 Nd4 10.Qd1 a6 0-1(73) Stockfish 1.6 JA 64bit_dec09 - FireBird 1.0 beta x64 mp, ? 2010.

[13] 10.Bxc6 bxc6 =-=(38) Stockfish 1.6 JA 64bit_dec09 - Rybka 3, ? 2010.

[14] 5.Qh5+ g6 6.Qh4 Bg7 7.dxe4 h6 8.Nf3 g5 9.Qh5+ Kf8 10.exd5 Nf6 =-=(117) Naum 4 MPx64 - Stockfish 1.6 JA 64bit_dec09, ? 2010.

[15] 8...Bf6 9.e5 Bxe5 10.Nc3 Bxc3+ 1-0(67) Naum 4 MPx64 - FireBird 1.0 beta x64 mp, ? 2010; 10...Bf6 0-1(63) Deep Fritz 11 - Rybka 3, ? 2010.
8...dxe4 9.Qxe4 Nc6 10.Bb5 Bf5 =-=(71) Zappa Mexico II - Deep Fritz 11, ? 2010.
8...Qe6 9.Nc3 dxe4 10.Nd2 e3 =-=(112) Deep Fritz 11 - Naum 4 MPx64, ? 2010.

[16] 10.exd5 Bf5 =-=(80) Naum 4 MPx64 - Zappa Mexico II, ? 2010; =-=(36) Deep Fritz 11 - Zappa Mexico II, ? 2010.

[17] 5...Nf6 6.dxe4 h6 7.Nb5 Qb6 8.e5 Ng4 9.Nh3 Nc6 10.Qxd5 a6 1-0(44) Zappa Mexico II - Stockfish 1.6 JA 64bit_dec09, ? 2010.

[18] 8.Bb5+ Nc6 9.Bxc6+ bxc6 10.Qe2 Qg6 =-=(77) Stockfish 1.6 JA 64bit_dec09 - Naum 4 MPx64, ? 2010.

[19] 10.g4 Be7 1-0(51) Deep Shredder 12 x64 FrGui - Zappa Mexico II, ? 2010; 10...Be6 0-1(66) Deep Shredder 12 x64 FrGui - FireBird 1.0 beta x64 mp, ? 2010.

[20] 8.Ncxd5 Nxd5 9.Qh5+ g6 10.Qxd5 e5 =-=(93) Zappa Mexico II - Naum 4 MPx64, ? 2010; 9...Kd8 10.Qxd5 Nb4 =-=(60) Naum 4 MPx64 - Deep Shredder 12 x64 FrGui, ? 2010.

[21] 10.Nxd1 Rh7 =-=(145) Deep Shredder 12 x64 FrGui - Naum 4 MPx64, ? 2010; 10...Rg8 =-=(133) Deep Shredder 12 x64 FrGui - Stockfish 1.6 JA 64bit_dec09, ? 2010.

[22] 10.Bd2 Qxd2+ 1-0(73) Stockfish 1.6 JA 64bit_dec09 - Zappa Mexico II, ? 2010.

[23] 10.Ne5+ Kd6 =-=(14) FireBird 1.0 beta x64 mp - Naum 4 MPx64, ? 2010; =-=(13) FireBird 1.0 beta x64 mp - Zappa Mexico II, ? 2010; 1-0(35) Rybka 3 - Naum 4 MPx64, ? 2010; 1-0(50) Rybka 3 - Zappa Mexico II, ? 2010; 10...Ke6 1-0(62) Rybka 3 - Deep Fritz 11, ? 2010; =-=(15) FireBird 1.0 beta x64 mp - Deep Fritz 11, ? 2010; 10...Kd8 =-=(16) Rybka 3 - Stockfish 1.6 JA 64bit_dec09, ? 2010.

Scores
Crosstable, Lisitsin

Code: Select all

 GGTJan10_4221_10+12_A03Lisitsin
                                  1  2  3  4  5  6  7  
1   Rybka 3                       ** 01 11 ½1 ½½ 11 11   9.5/12
2   FireBird 1.0 beta x64 mp      10 ** 10 0½ 11 ½½ 1½   7.0/12  38.00
3   Deep Shredder 12 x64 FrGui    00 01 ** ½½ ½½ 11 11   7.0/12  33.00
4   Naum 4 MPx64                  ½0 1½ ½½ ** ½½ ½1 ½½   6.5/12
5   Stockfish 1.6 JA 64bit_dec09  ½½ 00 ½½ ½½ ** ½0 10   4.5/12
6   Deep Fritz 11                 00 ½½ 00 ½0 ½1 ** ½½   4.0/12
7   Zappa Mexico II               00 0½ 00 ½½ 01 ½½ **   3.5/12
-----------------------------------------------------------------------
42 games

--------------------------------------------------------
Scores of all 7 gambits, Amar+Ware+Sokolsky+Swiss+From+Sturm+Lisitsin

Code: Select all

GGTJan 2010 40/20_20/10_10+12
                               Points
----------------------------------------------------
 1: Rybka 3                       58.0 / 84   
 2: FireBird 1.0 beta x64 mp      57.5 / 84   
 3: Stockfish 1.6 JA 64bit_dec09  38.5 / 84   
 4: Naum 4 MPx64                  37.0 / 84   
 5: Deep Shredder 12 x64 FrGui    37.0 / 84   
 6: Zappa Mexico II               36.0 / 84   
 7: Deep Fritz 11                 30.0 / 84 
----------------------------------------------------
294 games

Code: Select all

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws
  1 Rybka 3                        : 3198   60  58    84    69.0 %   3059   40.5 %
  2 FireBird 1.0 beta x64 mp       : 3194   62  60    84    68.5 %   3060   36.9 %
  3 Stockfish 1.6 JA 64bit_dec09   : 3054   55  55    84    45.8 %   3083   46.4 %
  4 Naum 4 MPx64                   : 3043   55  56    84    44.0 %   3085   45.2 %
  5 Deep Shredder 12 x64 FrGui     : 3043   65  65    84    44.0 %   3085   26.2 %
  6 Zappa Mexico II                : 3036   55  56    84    42.9 %   3086   45.2 %
  7 Deep Fritz 11                  : 2991   59  60    84    35.7 %   3093

Rybka has won now 5 gambits in a row and leads for the fist time with 1/2 point. Experienced engine match watchers know very well that such lucky streaks occur frequently, that's statistics..
All in all, the table is not a bad representation of the lists after only 14 games (rounds)each engine played against each other, isn't it? I don't understand why they need thousands of games to get it Well, perhaps someone will talk now about measurement scales, reliability, validity, significance, error margins and sample size!?

PGN-link:

http://www.file-upload.net/download-225 ... n.pgn.html

Next: Firebird wins Jänisch Gambit and takes back the lead

Test conditions
Starting positions: pool of 100 gambits sorted by Elo number.
Time Control: short tournament level with 40/20', 20/10', 10'+12''.
System: Intel Core i7 920, oc 3700 MHz, 6 GB RAM. Hyperthreading off, Turbo Mode off. Vista 64Bit
Parameter: 3 Threads, ponder off, 1,2 GB Hash, EGTB 3-4-5 (not used by Firebird and Stockfish, but sometimes used by the GUI, even for both engines)
Fritz11-GUI: remis never, resign late.

CU next round
Rainer Neuhäusler

[Albert Silver]

Rybka has won now 5 gambits in a row and leads for the fist time with 1/2 point. Experienced engine match watchers know very well that such lucky streaks occur frequently, that's statistics..

All in all, the table is not a bad representation of the lists after only 14 games (rounds)each engine played against each other, isn't it? I don't understand why they need thousands of games to get it Well, perhaps someone will talk now about measurement scales, reliability, validity, significance, error margins and sample size!?

I like the tournaments but don't understand your comments.

First you explain that Rybka won 5 gambits (60 games total) in a row through luck, then you claim that 84 games is enough to reliably measure an engine's strength. So according to you 60 games is still luck, but 84 games is skill?

[beram]

[quote="Rainer Marian"]+GGT+ Deep Rybka 3 wins Lisitsin Gambit and takes the lead+

.....
Next: Firebird wins Jänisch Gambit and takes back the lead


Does this quote above, means that this gambit round has finished yet ?

grts Bram

[rainhaus]

+GGT+ Deep Rybka 3 wins Lisitsin Gambit and takes the lead+
Next: Firebird wins Jänisch Gambit and takes back the lead

Does this quote above, means that this gambit round has finished yet ?
grts Bram

Yes, the Jaenisch is already played, but not yet tabulated and texted. Hopefully I will be able to do it soon.

cu RN

[rainhaus]

Rybka has won now 5 gambits in a row and leads for the fist time with 1/2 point. Experienced engine match watchers know very well that such lucky streaks occur frequently, that's statistics..

All in all, the table is not a bad representation of the lists after only 14 games (rounds)each engine played against each other, isn't it? I don't understand why they need thousands of games to get it Well, perhaps someone will talk now about measurement scales, reliability, validity, significance, error margins and sample size!?

I like the tournaments but don't understand your comments.

First you explain that Rybka won 5 gambits (60 games total) in a row through luck, then you claim that 84 games is enough to reliably measure an engine's strength. So according to you 60 games is still luck, but 84 games is skill?

-------------------------------------------------------------------------

A.Silver:I like the tournaments but don't understand your comments.

Nice, that you enjoy the GGT. I hope there are at least a few comments intelligible to you

A.Silver:First you explain that Rybka won 5 gambits (60 games total) in a row through luck.

Yes, on these four gambits Firebird scored with 39 points and rybka with 44,5 points. This is far away from any statistical significance. The very close results in winning a gambit or not speak the same language.

A. Silver: then you claim that 84 games is enough to reliably measure an engine's strength. So according to you 60 games is still luck, but 84 games is skill?

No, that's a short-circuit interpretation. I speak about a good correlation with the lists and not of reliable and valid results. The validity (not reliability) of my small sample is measured by the thousands of games by the lists. I borrow their true ranking and validate my own results with it. So is colloquially said, what testing validity does and how it is defined. Only the further results will show the stability of the current GGT ranking.
The twinkling smiley on the phrase "I don't understand why they need thousands of games to get it.." should point out an upside down irony by an impish reversing of the true relations between reference data and sample data. Ok, perhaps a little too much at the same time.
Recently I've read and old principle of journalism: Ironie, versteht der Leser nie. To get also a rhyme I'll try it with the following translation: Irony is not the reader's fantasy! Well, perhaps sometimes, if he wants

Thx you've talked with me about reliability, validity and sample size.
Cheers
Rainer Neuhäusler