Originally posted by cmsMasterI've been noticing this sort of thing for a while: when the evaluation changes dramatically for the worse, Fritz often will insert the text "gives away a clear win" regardless of whether the side moving was winning or not. It's a minor irritant, as the verbal annotations offer more comedy than information anyway (cats in trees, bulls on ice, etc.)
I had F9 analyze a game I had just finished on PC (after I annotated it myself first), and there was a strange glitch in the analysis. Look at it's comment/line for whites move 25. It says that white gives away a clear win, but after the line F9 gives black seems to be far ahead, and it's clearly not a win. On top of that while I was annotating After my ...[text shortened]... wever, after I played Rg8 Qh6 it gave black a -+ evaluation. Anyhow, can anybody explain it?
Here's a fairly recent example of the same sort of thing. In this case, the absurdity comes in the opening and Fritz finds the alternative offers equality.
[Event "Rated game, 15m + 1s"]
[Site "Café"]
[Date "2006.11.20"]
[Round "?"]
[White "Wulebgr"]
[Black "Sir Tommes"]
[Result "1/2-1/2"]
[ECO "D11"]
[WhiteElo "1702"]
[BlackElo "1608"]
[Annotator "Fritz 9 (15s)"]
[PlyCount "103"]
[EventDate "2006.11.20"]
[TimeControl "900+1"]
{D11: Slav Defence: 3 Nf3 sidelines and 3...Nf6 4 e3 Bg4} 1. d4 {4} d5 {4} 2.
c4 {1} c6 {2} 3. Nf3 {2} a6 {5} 4. g3 $146 {1} (4. Qc2 Nf6 5. Bf4 (5. e3 Bg4 6.
Ne5 Bh5 7. Bd3 e6 8. O-O Nfd7 9. Nxd7 Nxd7 10. Nc3 f5 11. Qb3 Rb8 12. Bd2 Bd6
13. cxd5 cxd5 14. Ne2 Qe7 15. a3 O-O 16. Bb4 Bxe2 17. Bxd6 Qxd6 18. Bxe2 Rfc8
19. Rfc1 Rc7 {
Ovsejevitsch,S (2493)-Papp,G (2362)/Budapest HUN 2003/The Week in Chess/1/2-1/2
}) 5... g6 (5... Bg4 6. Ne5 Bh5 7. Qb3 Ra7 8. Qh3 dxc4 9. Nc3 b5 10. g4 Bg6 11.
Bg2 Nd5 12. Nxd5 cxd5 13. g5 e6 14. Nxg6 Bb4+ 15. Kf1 fxg6 16. Qxe6+ Re7 17.
Qxd5 Rd7 18. Qe4+ Re7 19. Qa8 Bd6 20. Bxd6 {
Pelletier,Y (2624)-Slobodjan,R (2514)/Bischwiller FRA 2003/1-0 (40)}) 6. c5 Bg7
7. Nbd2 Nh5 8. Bg5 h6 9. Bh4 g5 10. Bg3 Nxg3 11. hxg3 Nd7 12. e3 g4 13. Nh4 e6
14. Be2 Qg5 15. b4 O-O 16. Nb3 f5 17. a4 Nf6 18. Bd3 Ne4 {
Gausel,E (2518)-Lyrberg,P (2432)/Skara SWE 2004/The Week in Chess 554/1-0 (78)}
) (4. cxd5 cxd5 5. Nc3 Nc6 6. Bf4 Bf5 7. e3 e6 8. Bd3 Bxd3 9. Qxd3 Bd6 {
1/2-1/2 Goczo,M (2222)-Martini,B (2278)/Budapest HUN 2003/The Week in Chess 438
}) 4... dxc4 {5} 5. Bg2 {4} b5 {4} (5... Nf6 6. O-O $14) 6. a4 {2} (6. O-O Nf6
$16) 6... Bb7 $14 {4} 7. O-O {3} e6 {11} 8. Ne5 {3} (8. b3 c5 9. axb5 axb5 10.
Rxa8 Bxa8 11. bxc4 bxc4 12. Qa4+ Nd7 $14) 8... Bd6 {13} (8... f6 9. Nf3 $11) 9.
axb5 {7} (9. b3 $5 cxb3 10. Nc3 $14) 9... axb5 $11 {10} 10. Rxa8 {6} Bxa8 {1}
11. Nc3 {12} Qb6 {25} (11... Ne7 12. Nf3 $11) 12. Be3 {8} (12. b3 $142 cxb3 13.
Bf4 $14) 12... b4 $4 {gives away a clear win 11} (12... Nf6 $142 $11 {
and Black can hope to live}) 13. Na4 {1} (13. Nxc4 $142 {ends the debate} Qd8
14. Nxd6+ Qxd6 15. Ne4 $18) 13... Qb5 $16 {10} 14. Qa1 {30} Bb7 {18} (14... f6
15. Nxc4 Qxc4 16. Nb6 $16) 15. Nc5 {5} (15. Nb6 $142 $1 Bxe5 16. dxe5 $18)
15... Bxc5 $14 {14} 16. dxc5 {White has the pair of bishops. 4} Ne7 {24} 17.
Rd1 {24} (17. Rc1 $5 {might be a viable alternative} Nf5 18. Nxc4 $14 (18. Rxc4
f6 19. Bd2 Na6 $14)) 17... O-O {9} 18. Nd7 {3} (18. Qc1 $5 {
should be considered} Ba6 19. Bh3 $15) 18... Nxd7 $17 {16} 19. Rxd7 {2} Nd5 {10
} (19... Nf5 $142 20. Bg5 Bc8 $17) 20. Qa7 $4 {19} (20. Bd4 $142 $15 {
was much better}) 20... Bc8 $19 {45} 21. Rd6 {40} Qb7 $4 {
not a good decision, because now the opponent is right back in the game 89} (
21... c3 $142 {and Black wins} 22. bxc3 Nxe3 23. cxb4 Nf5 $19 (23... Nxg2 $143
24. Qe7 h5 25. Rd8 Rxd8 26. Qxd8+ Kh7 27. Qxc8 $19)) 22. Qxb7 $15 {9} Bxb7 {2}
23. Bxd5 {1} (23. Bd2 $5 $15) 23... exd5 $17 {11} ({Inferior is} 23... cxd5 24.
c6 Bc8 25. c7 $14) 24. Rd7 {7} (24. Bd2 c3 25. bxc3 b3 $17) 24... Ba8 {11} (
24... Ba6 25. Rd6 Rc8 26. Bd4 $17) 25. Bf4 {4} f6 {28} 26. Bd6 {
A beautiful square! 5} (26. Kf1 $5 $17) 26... Re8 {4} 27. e3 $4 {
simply worsens the situation 6} (27. Kf1 b3 28. Ra7 c3 $17) 27... h6 {4} (27...
c3 {and Black can already relax} 28. bxc3 b3 29. g4 $19) 28. h4 $4 {
further deteriorates the position 2} (28. Ra7 b3 29. Ra6 Bb7 $17) 28... Kh7 {1}
(28... c3 $142 {keeps an even firmer grip} 29. bxc3 b3 30. g4 $19) 29. h5 $4 {
another bit of territory lost 1} (29. Ra7 $142 $17) 29... c3 $19 {17} 30. bxc3
{2} bxc3 {1} (30... b3 $142 {and Black can celebrate victory} 31. Kg2 b2 32.
Be7 $19) 31. Bf4 $4 {solves nothing 12} (31. Ra7 $142 $17) 31... c2 $19 {2} 32.
e4 {0} dxe4 {2} (32... Rxe4 $142 {and Black wins} 33. Bd2 Rd4 34. Be3 Rd1+ 35.
Kg2 $19) 33. Be3 $4 {
White has lost his nerve... understandable when you consider his position 6} (
33. Bc1 $142 $19) 33... Rb8 {11} 34. Kg2 {27} Rb1 {2} 35. Ra7 {2} (35. Rd4 {
does not improve anything} c1=Q 36. Bxc1 Rxc1 37. Rd7 $19) 35... c1=Q {8} (
35... Bb7 $142 {secures victory} 36. Ra4 c1=Q 37. Bxc1 Rxc1 38. Rxe4 Rxc5 $19)
36. Bxc1 {1} Rxc1 {1} 37. Rxa8 {1} Rxc5 {1} 38. Re8 {2} (38. Rc8 Rxh5 39. Rxc6
Re5 $19) 38... f5 {2} (38... Re5 39. Rc8 $19) 39. Re6 {3} (39. Rc8 Rc2 40. Kf1
Rc1+ 41. Kg2 Rc5 $19) 39... Rc2 {8} (39... Kg8 40. Kf1 $19) 40. Kf1 {5} (40.
Re5 g6 41. Re7+ Kg8 42. hxg6 Rd2 $17) 40... c5 {3} (40... Rc1+ 41. Ke2 Rh1 42.
Re5 Rxh5 43. Rc5 $19) 41. Ke1 {1} (41. Re7 Rc1+ 42. Kg2 Rc3 $17) 41... Rc3 {2}
(41... Rc1+ $142 42. Ke2 Rh1 $19) 42. g4 $4 {causes even greater problems 4} (
42. Re5 $142 $17) 42... fxg4 $19 {14} 43. Rxe4 {2} (43. Re5 $19 {
hardly improves anything}) 43... c4 {6} (43... g3 $142 {
finishes off the opponent} 44. fxg3 Rxg3 $19) 44. Rxg4 $15 {3} Rh3 {
Black threatens to win material: Rh3xh5 12} 45. Rxc4 {4} Rxh5 {
Black has a new protected passed pawn: h6 2} 46. Kf1 {2} Rf5 {3} 47. Kg2 {2}
Kg6 {2} 48. Rc6+ {5} Kg5 {3} 49. Kg3 {2} g6 {5} (49... Ra5 $5 $15) 50. f4+ $11
{12} Rxf4 {3} 51. Rxg6+ {Wulebgr offers a draw. . 1} Kxg6 {
Theme: Deflection from f4 11} 52. Kxf4 {
Sir Tommes bietet Remis (Lag: Av=0.80s, max=6.3s) 1} 1/2-1/2
Originally posted by torspoThis is what I was thinking, that Fritz thought there was a draw and the position was level
Engine just sees that position is repeated twice, that's why it gives it 0.00. If you put that position to the engine without previous moves, it won't do that and gives more accurate evaluation on position.
HTH
Do you think that was it cms?
Originally posted by Dies IraeYeah, that seems the only reasonable response, but it's still so weird that after the move is played F9 realizes it's clearly won for black.
This is what I was thinking, that Fritz thought there was a draw and the position was level
Do you think that was it cms?
In case you were interested. This is what Rybka had to say with 30s
[Event "Rated game, 15m + 0s"]
[Site "Main Playing Hall"]
[Date "2007.01.16"]
[Round "?"]
[White "Sandroalexis"]
[Black "CMSMaster"]
[Result "0-1"]
[ECO "B02"]
[WhiteElo "1119"]
[BlackElo "1262"]
[Annotator "Rybka 2.1c 32-bit (30s)"]
[PlyCount "68"]
[EventDate "2007.01.16"]
[EventType "rapid"]
{B02: Alekhine's Defence: Chase Variation and lines with early Nc3} 1. e4 Nf6
2. Nc3 d5 3. e5 Ne4 4. Qh5 $146 (4. Nge2 Bf5 5. Nd4 e6 6. Nxf5 exf5 7. Qf3 Qd7
8. d4 Bb4 9. Be2 Nc6 10. Qd3 Be7 11. O-O O-O-O 12. a3 f6 13. exf6 Bxf6 14. Be3
Rhe8 15. Qd1 Qe7 16. Nxe4 fxe4 17. b4 Qf7 18. c3 Kb8 {
Kausilaite,V (2000)-Sarkisova,N (2000)/Moscow 1979/MCD/0-1 (58)}) (4. f3 Nxc3
5. dxc3 c5 6. f4 Nc6 7. Nf3 Bf5 8. Bd3 e6 9. Bxf5 exf5 10. Qd3 Qd7 11. Be3 Be7
12. O-O-O O-O-O 13. Rd2 Qe6 14. b3 Rd7 15. Rhd1 Rhd8 16. Bf2 g6 17. Bh4 Bxh4
18. Nxh4 Nb8 {Bogumil,P (2355)-Donchev,D (2485)/Pardubice 1993/TD/1/2-1/2 (49)}
) (4. Bd3 Nxc3 5. bxc3 c5 6. Bb2 Nc6 7. c4 Nb4 8. Qf3 e6 9. cxd5 exd5 10. Ba3
Nxd3+ 11. Qxd3 c4 12. Qc3 d4 13. Qf3 Be7 14. Bxe7 Qxe7 15. Qd5 d3 16. cxd3 cxd3
17. Qxd3 Qxe5+ 18. Kd1 Qxa1+ {
Kers,K (1128)-Schakel,M (1579)/Hengelo 2005/CBM 107 ext/0-1}) (4. Nb1 d4 5. f4
Nc6 6. Nf3 e6 7. d3 Nc5 8. c4 b6 9. Na3 Bb7 10. Nc2 a5 11. Be2 Nb4 12. O-O Nxc2
13. Qxc2 Qd7 14. b3 Na6 15. a3 f5 16. exf6 gxf6 17. Nh4 O-O-O 18. Bf3 Rg8 {
Hartmann,W (2240)-Vogler,T/Germany 1981/GER-chT/1-0 (50)}) (4. f4 Bf5 5. Nce2
e6 6. d3 Nc5 7. Ng3 Bg6 8. Be2 Be7 9. Nf3 Nc6 10. a3 a5 11. Be3 d4 12. Bf2 a4
13. Nd2 Qd7 14. Bf3 Na5 15. O-O Rb8 16. Ne2 Nc6 17. Bxc6 Qxc6 18. Nxd4 Qa6 {
Kessler,A-Meyer,I/Schweich 1990/GER/1/2-1/2 (90)}) 4... Nxc3 5. bxc3 e6 6. Bd3
Be7 7. Nf3 Nc6 8. Bb5 Bd7 9. Bxc6 Bxc6 10. d4 Bb5 {
Black traps the enemy king in the center} 11. Rb1 {
White threatens to win material: Rb1xb5} Ba6 12. Ng5 Bxg5 (12... g6 13. Qh6 Bf8
14. Qh3 $15) 13. Bxg5 {Opposite coloured bishops appeared} Qd7 14. Rd1 (14. Rb4
O-O $11) 14... O-O (14... Qa4 15. Kd2 $15) 15. Qg4 {Black king safety improved}
(15. Bf6 gxf6 16. Qh6 Qc6 17. exf6 Qxc3+ 18. Rd2 Qa1+ 19. Rd1 Qc3+ 20. Rd2 Qa1+
21. Rd1 Qc3+ $11) 15... Kh8 $15 16. Bf6 $2 (16. Ra1 $142 $15 {
and White hangs on}) 16... gxf6 $19 17. exf6 (17. Qh4 Rg8 $19) 17... Rg8 18.
Qh5 (18. Qf3 Rae8 $19) 18... Rg6 19. Rd3 (19. g3 Rag8 20. f4 $19) 19... Bxd3
20. cxd3 Rxf6 (20... Qb5 $5 {and Black can already relax} 21. O-O $19) 21. O-O
Rf5 22. Qh6 Rg8 23. Rb1 Rg6 (23... Qc6 $142 {
it becomes clear that Black will call all the shots} 24. Qd2 Qa6 $19) 24. Qf8+
$11 Rg8 {Black threatens to win material: Rg8xf8} 25. Qb4 $4 {
letting the wind out of his own sails} (25. Qh6 Qc6 26. Qd2 $19) 25... b6 (
25... Rh5 26. Qxb7 e5 27. g3 $19) 26. Qa3 (26. Qb2 {does not help much} e5 $19)
26... Rfg5 (26... e5 {keeps an even firmer grip} 27. g3 $19) 27. g3 f5 (27...
a5 {seems even better} 28. Qc1 $19) 28. Qc1 (28. Qxa7 {doesn't do any good} Qc6
29. Qa3 f4 $19) 28... Qf7 29. Qf4 Rg4 30. Qe5+ R8g7 31. Qe3 (31. Re1 $19 {
is the last straw}) 31... f4 32. Qf3 (32. Qe1 {the last chance for counterplay}
fxg3 33. fxg3 $19) 32... fxg3 33. Qxf7 gxh2+ 34. Kxh2 (34. Kh1 {
there is nothing else anyway} Rxf7 35. Rb2 $19) 34... Rh4# 0-1
Actually it 'understands' that white is lost, that' why repetition has better evaluation than any other move. When Qh6 move is played, evaluation is still the same. Black's best move is not Rg6 and F9 sees that.
To my understanding, your engine is giving exactly the right evaluation. It's not the way humans will play that position but ...
Originally posted by torspoThis is what I thought at first, and it seems to be related to the answer, but the mysterious thing is that Fritz claims that "white gave up a clearly winning position" instead of saying that "white missed the draw"
Actually it 'understands' that white is lost, that' why repetition has better evaluation than any other move. When Qh6 move is played, evaluation is still the same. Black's best move is not Rg6 and F9 sees that.
To my understanding, your engine is giving exactly the right evaluation. It's not the way humans will play that position but ...
Fritz goes bananas if you repeat moves.
It has to do with how it detects repetition. If you calculate 2 repetitions as draw you can do a whole lot of cutoffs in the search.
But you'd think they could fix that glitch atleast in the analysis and 'advice'.
If you have hints turned on and you repeat a move it always warns you that you made a huge mistake (if you were ahead) even though the position is not drawn yet.
About the "clearly winning" I am not sure why it would say that, most likely it only has a check for losing x amount of evaluation and puts in that line if you do. But it is probably considered a bug.
(sorry for repeating some that was already said 🙂