if assuming that sharpness is better chess.
I have a naive question. Is there any relation between trappy and sharp?
I have a naive question. Is there any relation between trappy and sharp?
I like the concept of a sharpness number for a given chess position. Edge cases:
0. A position with 100% draw chances has zero sharpness.
1. A position with 100% winning chances for one player has zero sharpness.
2. A position with 50% winning chances for both players has sharpness 1.0.
A function that does this is the standard deviation (with 1 point for W, 0 for D, -1 for L):
Normalize so that W+D+L=1
sd = W+L - (W-L)^2
Example 0: D=1 then W=L=0 and sd=0
Example 1: W=1 then L=0 and sd=0
Example 2: D=0,W=L=0.5 then sd=1
sd is the non-drawing chance minus the winning chance difference squared. Positions with higher drawing chances have lower (W+L) and so lower sd. One-sided positions have higher (W-L)^2 and so lower sd.
0. A position with 100% draw chances has zero sharpness.
1. A position with 100% winning chances for one player has zero sharpness.
2. A position with 50% winning chances for both players has sharpness 1.0.
A function that does this is the standard deviation (with 1 point for W, 0 for D, -1 for L):
Normalize so that W+D+L=1
sd = W+L - (W-L)^2
Example 0: D=1 then W=L=0 and sd=0
Example 1: W=1 then L=0 and sd=0
Example 2: D=0,W=L=0.5 then sd=1
sd is the non-drawing chance minus the winning chance difference squared. Positions with higher drawing chances have lower (W+L) and so lower sd. One-sided positions have higher (W-L)^2 and so lower sd.
Already years before there was these superhuman engines, I was trying to get to such sharpness measure, to identify players worth looking into - that meant: likeTal, like Shirov.
I was coming up with a very simple measure:
- Look at all games of a player
- Look at all capturing moves in those games (also capturing moves by the opponent)
- How many % of capturing moves are on the same square as the previous capture?
That leads to a score between 0% and 100%, and I found it quite indicative.
Basically this is measuring frequency of zwischenzugs (although blind for other zwischenzugs like check), and ignoring long (endgame) phases where not much is happening.
I was coming up with a very simple measure:
- Look at all games of a player
- Look at all capturing moves in those games (also capturing moves by the opponent)
- How many % of capturing moves are on the same square as the previous capture?
That leads to a score between 0% and 100%, and I found it quite indicative.
Basically this is measuring frequency of zwischenzugs (although blind for other zwischenzugs like check), and ignoring long (endgame) phases where not much is happening.
The repeat capture measure can be easily calculated directly from a PGN file. But the difference between games in this metric is subtle.
Sharpness in chess is not well defined. Maybe complicated is a better term. There are simple positions and complicated positions. A player wants to make the position complicated for their opponent, while keeping it simple for themself. Complicated positions have more chances of going wrong, and so a wider dispersion of results.
=== some games ===
World Chess Championship 2023 Game 2: Ding–Nepomniachtchi
Nepo quickly won this game, and so I would expect a high sharpness score.
12 captures
6 on different square than previous capture
repeat percentage = 6/11 = 55%
dxc4 Bxc4 Nxf6+ gxf6 Rxg5 Nxg5 Nxd4 Nxc2 Rxc2 Bxe4 bxc3 bxc3
1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. h3 dxc4 5. e3 c5 6. Bxc4 a6 7. 0-0 Nc6 8. Nc3 b5 9. Bd3 Bb7 10. a4 b4 11. Ne4 Na5 12. Nxf6+ gxf6 13. e4 c4 14. Bc2 Qc7 15. Bd2 Rg8 16. Rc1 0-0-0 17. Bd3 Kb8 18. Re1 f5 19. Bc2 Nc6 20. Bg5 Rxg5 21. Nxg5 Nxd4 22. Qh5 f6 23. Nf3 Nxc2 24. Rxc2 Bxe4 25. Rd2 Bd6 26. Kh1 c3 27. bxc3 bxc3 28. Rd4 c2 29. Qh6 e5 0–1
[Event "World Championship Candidates"][Round "14.1"]
[White "Hikaru Nakamura"][Black "D Gukesh"]
This was a quiet game, and so I would expect a low sharpness score.
30 captures
15 on different square than previous capture
repeat percentage = 15/29 = 52%
dxc4 Bxc4 cxd4 exd4 bxa3 bxa3 Nxc3 Bxc3 Bxf3 Qxf3 Qxd4 Qxb5 Bxb5 Rxc4 Bxc4 Bxd4 Rxd4 Rxh7 fxe4 fxe4 gxh4 gxh4 Bxh4 Rxa5 Rxh5 Rxa8 Rxa8 Rxe3 Bxe3 Kxe3
1. d4 d5 2. c4 dxc4 3. Nf3 Nf6 4. e3 e6 5. Bxc4 Be7 6. Nc3 c5 7. a3 O-O 8. O-O a6 9. Qe2 b5 10. Ba2 cxd4 11. exd4 b4 12. Ne4 bxa3 13. bxa3 Bb7 14. Nc3 Nd5 15. Bd2 Nxc3 16. Bxc3 Nd7 17. Bb1 Rc8 18. Bd2 Nf6 19. Bd3 Bxf3 20. Qxf3 Qxd4 21. Rfd1 Qa4 22. Qb7 Bc5 23. Be1 a5 24. Qb5 Qxb5 25. Bxb5 Bb6 26. Rac1 Nd5 27. Kf1 Rfd8 28. a4 Kf8 29. g3 Ke7 30. Ke2 f5 31. Rc4 Rxc4 32. Bxc4 Nb4 33. Rb1 Bc5 34. Rc1 Nc6 35. Bc3 g5 36. Bb5 Ba3 37. Rc2 Nd4+ 38. Bxd4 Rxd4 39. Rc7+ Kf6 40. Rxh7 Bb4 41. Bd3 e5 42. f3 e4 43. fxe4 fxe4 44. Bb5 Be7 45. h4 gxh4 46. gxh4 Rb4 47. Rh5 Kg7 48. Bd7 Rd4 49. Be8 Bd8 50. Rf5 Bxh4 51. Rxa5 Kf6 52. Ra8 Bg5 53. Bh5 Rd2+ 54. Kf1 Ra2 55. a5 Ra1+ 56. Kg2 Ra2+ 57. Kf1 Ra1+ 58. Ke2 Ra2+ 59. Kd1 Rd2+ 60. Ke1 Rd5 61. Ke2 Bd2 62. a6 Rxh5 63. a7 Ra5 64. Rf8+ Ke5 65. a8=Q Rxa8 66. Rxa8 e3 67. Ra4 Kd5 68. Kd3 Kc5 69. Re4 Kd5 70. Rxe3 Bxe3 71. Kxe3 1/2-1/2
[Event "World Championship Candidates"][Round "14.2"]
[White "Caruana, Fabiano"][Black "Nepomniachtchi, Ian"] (25 captures)
Nepo created confusion in a winning position for Fabi, and so I would expect a high sharpness score.
25 captures
15 on different square than previous capture
repeat percentage = 15/24 = 63%
cxd5 exd5 Nxg3 hxg3 Bxc3 Qxc3 Bxf3 gxf3 dxe4 Bxe4 gxf4 dxc6 Rxd7 cxd7 Nxd7 Qxf4+ Qxf7 Rxg5 Qxg5 Qxh7+ Nxe4 Qxa4+ Qxe4+ Qxf8 Qxf8
1. d4 Nf6 2. c4 e6 3. Nc3 d5 4. cxd5 exd5 5. Bg5 Bb4 6. e3 h6 7. Bh4 g5 8. Bg3 Ne4 9. Qc2 h5 10. f3 Nxg3 11. hxg3 Be6 12. Bd3 c6 13. O-O-O Nd7 14. f4 Bg4 15. Nf3 Qe7 16. e4 Bxc3 17. Qxc3 Bxf3 18. gxf3 dxe4 19. Bxe4 gxf4 20. g4 O-O-O 21. d5 h4 22. dxc6 Nc5 23. Bf5+ Kb8 24. Kb1 b6 25. Rd7 Rxd7 26. cxd7 Rd8 27. Qd4 Nxd7 28. Rd1 Qc5 29. Qxf4+ Qc7 30. Qd2 h3 31. Be4 a5 32. Qd5 Ka7 33. Qxf7 h2 34. Qh7 Kb8 35. a3 Qe5 36. Qh6 Qc7 37. g5 Rg8 38. Rh1 Nc5 39. Bh7 Rxg5 40. Qxg5 Qxh7+ 41. Ka1 Qc2 42. Qg8+ Ka7 43. Ka2 a4 44. f4 Nb3 45. Qg7+ Ka6 46. Qc3 Qg2 47. Qc4+ Kb7 48. Re1 Nc5 49. Qf1 Qd5+ 50. Kb1 Qf5+ 51. Ka1 Qc2 52. f5 Nd3 53. Rb1 Nc5 54. f6 Nb3+ 55. Ka2 Nd2 56. Qh1+ Ka7 57. Rc1 Qb3+ 58. Ka1 Qe6 59. Rc7+ Ka6 60. f7 Nb3+ 61. Kb1 Qf5+ 62. Ka2 Nc5 63. Qa8+ Kb5 64. Qc6+ Ka6 65. Qa8+ Kb5 66. Qc6+ Ka6 67. Re7 Qf1 68. Qa8+ Kb5 69. Qe8+ Ka6 70. Qa8+ Kb5 71. Qe8+ Ka6 72. Re4 Nxe4 73. Qxa4+ Kb7 74. Qxe4+ Ka7 75. Qa4+ Kb7 76. Qd7+ Ka6 77. Qc8+ Ka7 78. f8=Q Qxf8 79. Qxf8 h1=Q 80. a4 Qd5+ 81. Ka3 Qd3+ 82. b3 Qd4 83. Qf7+ Kb8 84. Qe8+ Ka7 85. Qe7+ Kb8 86. Qe1 Qd6+ 87. Qb4 Qf6 88. Qd2 Qe7+ 89. b4 Qe6 90. Qd3 Ka7 91. Qc3 Kb8 92. Qb3 Qf6 93. Qe3 Qd6 94. Kb3 Qd5+ 95. Kb2 Qd6 96. Qe8+ Ka7 97. Qf7+ Kb8 98. Kc3 Qe5+ 99. Kd3 Qd6+ 100. Kc4 Qc6+ 101. Kb3 Qd6 102. Qf3 Qd4 103. Qe2 Qd5+ 104. Ka3 Qd6 105. Qb5 Qd4 106. Kb3 Ka7 107. Qe2 Qd5+ 108. Ka3 Qd4 109. Kb3 1/2-1/2
Sharpness in chess is not well defined. Maybe complicated is a better term. There are simple positions and complicated positions. A player wants to make the position complicated for their opponent, while keeping it simple for themself. Complicated positions have more chances of going wrong, and so a wider dispersion of results.
=== some games ===
World Chess Championship 2023 Game 2: Ding–Nepomniachtchi
Nepo quickly won this game, and so I would expect a high sharpness score.
12 captures
6 on different square than previous capture
repeat percentage = 6/11 = 55%
dxc4 Bxc4 Nxf6+ gxf6 Rxg5 Nxg5 Nxd4 Nxc2 Rxc2 Bxe4 bxc3 bxc3
1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. h3 dxc4 5. e3 c5 6. Bxc4 a6 7. 0-0 Nc6 8. Nc3 b5 9. Bd3 Bb7 10. a4 b4 11. Ne4 Na5 12. Nxf6+ gxf6 13. e4 c4 14. Bc2 Qc7 15. Bd2 Rg8 16. Rc1 0-0-0 17. Bd3 Kb8 18. Re1 f5 19. Bc2 Nc6 20. Bg5 Rxg5 21. Nxg5 Nxd4 22. Qh5 f6 23. Nf3 Nxc2 24. Rxc2 Bxe4 25. Rd2 Bd6 26. Kh1 c3 27. bxc3 bxc3 28. Rd4 c2 29. Qh6 e5 0–1
[Event "World Championship Candidates"][Round "14.1"]
[White "Hikaru Nakamura"][Black "D Gukesh"]
This was a quiet game, and so I would expect a low sharpness score.
30 captures
15 on different square than previous capture
repeat percentage = 15/29 = 52%
dxc4 Bxc4 cxd4 exd4 bxa3 bxa3 Nxc3 Bxc3 Bxf3 Qxf3 Qxd4 Qxb5 Bxb5 Rxc4 Bxc4 Bxd4 Rxd4 Rxh7 fxe4 fxe4 gxh4 gxh4 Bxh4 Rxa5 Rxh5 Rxa8 Rxa8 Rxe3 Bxe3 Kxe3
1. d4 d5 2. c4 dxc4 3. Nf3 Nf6 4. e3 e6 5. Bxc4 Be7 6. Nc3 c5 7. a3 O-O 8. O-O a6 9. Qe2 b5 10. Ba2 cxd4 11. exd4 b4 12. Ne4 bxa3 13. bxa3 Bb7 14. Nc3 Nd5 15. Bd2 Nxc3 16. Bxc3 Nd7 17. Bb1 Rc8 18. Bd2 Nf6 19. Bd3 Bxf3 20. Qxf3 Qxd4 21. Rfd1 Qa4 22. Qb7 Bc5 23. Be1 a5 24. Qb5 Qxb5 25. Bxb5 Bb6 26. Rac1 Nd5 27. Kf1 Rfd8 28. a4 Kf8 29. g3 Ke7 30. Ke2 f5 31. Rc4 Rxc4 32. Bxc4 Nb4 33. Rb1 Bc5 34. Rc1 Nc6 35. Bc3 g5 36. Bb5 Ba3 37. Rc2 Nd4+ 38. Bxd4 Rxd4 39. Rc7+ Kf6 40. Rxh7 Bb4 41. Bd3 e5 42. f3 e4 43. fxe4 fxe4 44. Bb5 Be7 45. h4 gxh4 46. gxh4 Rb4 47. Rh5 Kg7 48. Bd7 Rd4 49. Be8 Bd8 50. Rf5 Bxh4 51. Rxa5 Kf6 52. Ra8 Bg5 53. Bh5 Rd2+ 54. Kf1 Ra2 55. a5 Ra1+ 56. Kg2 Ra2+ 57. Kf1 Ra1+ 58. Ke2 Ra2+ 59. Kd1 Rd2+ 60. Ke1 Rd5 61. Ke2 Bd2 62. a6 Rxh5 63. a7 Ra5 64. Rf8+ Ke5 65. a8=Q Rxa8 66. Rxa8 e3 67. Ra4 Kd5 68. Kd3 Kc5 69. Re4 Kd5 70. Rxe3 Bxe3 71. Kxe3 1/2-1/2
[Event "World Championship Candidates"][Round "14.2"]
[White "Caruana, Fabiano"][Black "Nepomniachtchi, Ian"] (25 captures)
Nepo created confusion in a winning position for Fabi, and so I would expect a high sharpness score.
25 captures
15 on different square than previous capture
repeat percentage = 15/24 = 63%
cxd5 exd5 Nxg3 hxg3 Bxc3 Qxc3 Bxf3 gxf3 dxe4 Bxe4 gxf4 dxc6 Rxd7 cxd7 Nxd7 Qxf4+ Qxf7 Rxg5 Qxg5 Qxh7+ Nxe4 Qxa4+ Qxe4+ Qxf8 Qxf8
1. d4 Nf6 2. c4 e6 3. Nc3 d5 4. cxd5 exd5 5. Bg5 Bb4 6. e3 h6 7. Bh4 g5 8. Bg3 Ne4 9. Qc2 h5 10. f3 Nxg3 11. hxg3 Be6 12. Bd3 c6 13. O-O-O Nd7 14. f4 Bg4 15. Nf3 Qe7 16. e4 Bxc3 17. Qxc3 Bxf3 18. gxf3 dxe4 19. Bxe4 gxf4 20. g4 O-O-O 21. d5 h4 22. dxc6 Nc5 23. Bf5+ Kb8 24. Kb1 b6 25. Rd7 Rxd7 26. cxd7 Rd8 27. Qd4 Nxd7 28. Rd1 Qc5 29. Qxf4+ Qc7 30. Qd2 h3 31. Be4 a5 32. Qd5 Ka7 33. Qxf7 h2 34. Qh7 Kb8 35. a3 Qe5 36. Qh6 Qc7 37. g5 Rg8 38. Rh1 Nc5 39. Bh7 Rxg5 40. Qxg5 Qxh7+ 41. Ka1 Qc2 42. Qg8+ Ka7 43. Ka2 a4 44. f4 Nb3 45. Qg7+ Ka6 46. Qc3 Qg2 47. Qc4+ Kb7 48. Re1 Nc5 49. Qf1 Qd5+ 50. Kb1 Qf5+ 51. Ka1 Qc2 52. f5 Nd3 53. Rb1 Nc5 54. f6 Nb3+ 55. Ka2 Nd2 56. Qh1+ Ka7 57. Rc1 Qb3+ 58. Ka1 Qe6 59. Rc7+ Ka6 60. f7 Nb3+ 61. Kb1 Qf5+ 62. Ka2 Nc5 63. Qa8+ Kb5 64. Qc6+ Ka6 65. Qa8+ Kb5 66. Qc6+ Ka6 67. Re7 Qf1 68. Qa8+ Kb5 69. Qe8+ Ka6 70. Qa8+ Kb5 71. Qe8+ Ka6 72. Re4 Nxe4 73. Qxa4+ Kb7 74. Qxe4+ Ka7 75. Qa4+ Kb7 76. Qd7+ Ka6 77. Qc8+ Ka7 78. f8=Q Qxf8 79. Qxf8 h1=Q 80. a4 Qd5+ 81. Ka3 Qd3+ 82. b3 Qd4 83. Qf7+ Kb8 84. Qe8+ Ka7 85. Qe7+ Kb8 86. Qe1 Qd6+ 87. Qb4 Qf6 88. Qd2 Qe7+ 89. b4 Qe6 90. Qd3 Ka7 91. Qc3 Kb8 92. Qb3 Qf6 93. Qe3 Qd6 94. Kb3 Qd5+ 95. Kb2 Qd6 96. Qe8+ Ka7 97. Qf7+ Kb8 98. Kc3 Qe5+ 99. Kd3 Qd6+ 100. Kc4 Qc6+ 101. Kb3 Qd6 102. Qf3 Qd4 103. Qe2 Qd5+ 104. Ka3 Qd6 105. Qb5 Qd4 106. Kb3 Ka7 107. Qe2 Qd5+ 108. Ka3 Qd4 109. Kb3 1/2-1/2
Self play games would be more accurate.
so for some it is about the repeat capture (the explosion notion) and other it is about room to play good enough moves, versus being on the tip of mountain or ridge, with no plateau in foresight?
maybe those might overlap, or one could restrict the notion to be the intersection of those semantic fields (Venn diagram potatoes, if one needs me to draw that, I will).
maybe those might overlap, or one could restrict the notion to be the intersection of those semantic fields (Venn diagram potatoes, if one needs me to draw that, I will).
It's easier to define what is NOT sharp. It's when a chess player is NOT prepared with a response to a move. A response to a move is where you will see the preparedness and how sharp they are. If white has the initiative, than black must play sharp.
When the material is harassed or attacked, responding well is playing sharp. A player that has the initiative must be harassing or attacking the material of the opponent and the opponent must be able to do counter-measures to correct the threat. If the opponent is playing sharp, they are avoiding losses by not simplifying a position when they don't have the advantage to do so.
It's not the player that is harassing or attacking material that is playing sharp,they have the initiate. It's the skilled defender that must play sharp. This is where the definitions of sharp and initiative separate. If it was to be given a fractional value, the initiative would be the denominator and the sharpness value the numerator. How well we respond to a threat depends on many factors, but being keen minded, acute and skilled surely helps.
When the material is harassed or attacked, responding well is playing sharp. A player that has the initiative must be harassing or attacking the material of the opponent and the opponent must be able to do counter-measures to correct the threat. If the opponent is playing sharp, they are avoiding losses by not simplifying a position when they don't have the advantage to do so.
It's not the player that is harassing or attacking material that is playing sharp,they have the initiate. It's the skilled defender that must play sharp. This is where the definitions of sharp and initiative separate. If it was to be given a fractional value, the initiative would be the denominator and the sharpness value the numerator. How well we respond to a threat depends on many factors, but being keen minded, acute and skilled surely helps.