SF6 moves comparison: Event Vugar Gashimov Mem 2015

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

Dirt wrote:
Ferdy wrote:1. When the search score of the position is already outside of +/-2 pawns as analyzed by the engine, this position is not included.
So, if a player has a winning position and makes a mistake that results in an even score, it doesn't count as an error?
I am revising the program to include this in the error calculation.
Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

Vinvin wrote:Is it possible to get a list of the 20 biggest blunders ?
I can just output the epd given a threshold of say 1/2 or more pawn worst than the best.
Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Dirt »

Ferdy wrote:I am revising the program to include this in the error calculation.
Good. It just seemed like an odd choice.
Deasil is the right way to go.
Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

Tournament and players compared. Just a quick 1sec/pos, see also the notes. Sample epd files are attached.
If move is not best, epd positions are saved per player and per amount of error. These blunder positions saved are in append mode, so be sure to delete those if you run same epd. The summary is append mode also, the output below is an example. There is also a log output, for checking, this is overwrite mode, be sure to backup if you want to review previous results.

Code: Select all

Moves Error calculation based from Stockfish 6 64 POPCNT and players from
Event 2nd Shamkir Chess 2015, after round 5

Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Engine id name Stockfish 6 64 POPCNT
Hash 128 MB, Threads 1, MoveTime 1.0 sec/pos

 Rank                       Players  AveDepth  TotalPos  AveError  Err_3.0P  Err_2.0P  Err_1.0P  Err_0.5P    W    L    D  Games  Score  SRate

    1            Anand, Viswanathan        18       166      3.97         0         0         1         2    1    0    4      5    3.0   60.0%
    2                 Mamedov, Rauf        17       174      4.05         0         0         0         2    0    1    4      5    2.0   40.0%
    3               Carlsen, Magnus        18       171      4.26         0         0         0         1    3    0    2      5    4.0   80.0%
    4                    So, Wesley        16       147      5.25         0         0         2         4    3    1    1      5    3.5   70.0%
    5              Caruana, Fabiano        18       192      5.93         0         0         0         5    0    1    4      5    2.0   40.0%
    6        Mamedyarov, Shakhriyar        17       213      6.95         1         0         1         5    1    1    3      5    2.5   50.0%
    7       Vachier-Lagrave, Maxime        18       159      7.30         1         0         0         3    0    1    4      5    2.0   40.0%
    8             Kramnik, Vladimir        17       173      8.86         0         0         0         5    1    1    3      5    2.5   50.0%
    9                   Giri, Anish        16       154     10.71         0         1         1         2    0    1    4      5    2.0   40.0%
   10                Adams, Michael        16       134     16.30         0         0         1         4    0    2    3      5    1.5   30.0%

Overall Average Error: 7.36

Code: Select all

Moves Error calculation based from Stockfish 6 64 POPCNT and players from
Event New York 1927, after round 1

Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Engine id name Stockfish 6 64 POPCNT
Hash 128 MB, Threads 1, MoveTime 1.0 sec/pos

 Rank                       Players  AveDepth  TotalPos  AveError  Err_3.0P  Err_2.0P  Err_1.0P  Err_0.5P    W    L    D  Games  Score  SRate

    1         Capablanca, Jose Raul        17       147      4.74         0         0         0         2    3    0    2      5    4.0   80.0%
    2            Nimzowitsch, Aaron        19       149      9.87         0         1         0         4    3    1    1      5    3.5   70.0%
    3             Spielmann, Rudolf        17       155     13.86         0         0         3         8    0    2    3      5    1.5   30.0%
    4              Vidmar, Milan Sr        17       119     14.83         0         1         0        11    0    1    4      5    2.0   40.0%
    5           Alekhine, Alexander        17       166     16.51         0         0         2         9    1    1    3      5    2.5   50.0%
    6         Marshall, Frank James        18       147     28.48         1         1         2         9    0    2    3      5    1.5   30.0%

Overall Average Error: 14.71

Code: Select all

Legend:
Rank     - ranking based on low average error
AveDepth - average depth reached by the engine in analyzing those positions
AveError - average difference between the engine and player score, in centipawn (1p = 100)
Err_3.0P - count of positions where player has blundered by 3 or more pawns
Err_2.0P - count of positions where player has blundered by 2 to 2.9 pawns
Err_1.0P - count of positions where player has blundered by 1 to 1.9 pawns
Err_0.5P - count of positions where player has blundered by 0.5 to 0.9 pawns

Notes:
1. Analysis of positions start at move 12.
2. If best score of engine is already above 3 pawns and the score of player move
   is also above 3 pawns then error for this position is set to zero. The more cases
   like this only proves that that player is able to maintain its advantage and this
   will minimize its average error.
   On the other hand if the best score of engine is already below 3 pawns, no further analysis
   will be done, and this position will not be included in error calculation.
3. When a player move is a blunder that brings down the score below 3 pawns worst,
   error for this position is set to 6 pawns in the calculation. The reason being once
   the opponent found the refutation, the succeeding positions will no longer be evaluated.
   It is just fair to penalize this heavily to also reflect on the average error which
   should be high for players that have more game loses or fatal blunders.
Sample blunder @ 3 pawns or more.
Mamedyarov, Shakhriyar_blunder3P.epd
br2qr1k/6bp/1N1Pp1p1/2Q1np2/4p3/2B1P3/4BPPP/2R2RK1 b - - fmvn 26; sm "e5d7"; c3 "Result 1-0"; c2 "Carlsen, Magnus / Mamedyarov, Shakhriyar"; c1 "Round 2"; c0 "Event 2nd Shamkir Chess 2015"; hmvc 10; c5 1; c4 1648;
[d]br2qr1k/6bp/1N1Pp1p1/2Q1np2/4p3/2B1P3/4BPPP/2R2RK1 b - -
In the game he played e5d7.

Download:
http://www.mediafire.com/download/xqfte ... orV8.2.rar
Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

After round 7 with improvement on determining move average error, see note 5.
New is column EECnt for Enginer Error Count, these are position counts
where player move score is better than the engine.

Code: Select all

Move Error Evaluator v8.5

Moves Error calculation based from Stockfish 180415 64 POPCNT and players from
Event Vugar Gashimov Mem 2015, after round 7

Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Engine id name Stockfish 180415 64 POPCNT
Hash 128 MB, Threads 1, MoveTime 1.0 sec/pos

 Rank                       Players  AveDepth  TotalPos  AveError  E3.0P  E2.0P  E1.0P  E0.5P  EECnt   EERate    W    L    D  Games  Score  SRate

    1                       Anand,V        18       223      8.40      0      0      2      6     24   10.76%    2    0    5      7    4.5   64.3%
    2                     Carlsen,M        20       235      8.47      0      0      2      2     16    6.81%    4    0    3      7    5.5   78.6%
    3                  Mamedyarov,S        18       260     10.77      1      0      3      2     23    8.85%    1    1    5      7    3.5   50.0%
    4                   Mamedov,Rau        18       212     11.00      0      0      1      1     18    8.49%    0    1    6      7    3.0   42.9%
    5                     Caruana,F        20       270     13.71      0      0      0      8     23    8.52%    2    1    4      7    4.0   57.1%
    6             Vachier Lagrave,M        20       241     14.88      1      0      1      3     22    9.13%    0    1    6      7    3.0   42.9%
    7                        Giri,A        18       214     20.18      0      1      0      6     11    5.14%    0    1    6      7    3.0   42.9%
    8                          So,W        18       208     22.63      0      0      1      8     15    7.21%    3    2    2      7    4.0   57.1%
    9                      Adams,Mi        18       215     26.31      0      0      4      6     16    7.44%    0    3    4      7    2.0   28.6%
   10                     Kramnik,V        18       238     30.85      0      0      3      9     23    9.66%    1    3    3      7    2.5   35.7%

Overall Average Error: 16.72

Legend:
Rank     - ranking based on low average error
AveDepth - average depth reached by the engine in analyzing those positions
AveError - average difference between the engine and player score, in centipawn (1p = 100)
E3.0P    - count of positions where player has blundered by 3 or more pawns
E2.0P    - count of positions where player has blundered by 2 to 2.9 pawns
E1.0P    - count of positions where player has blundered by 1 to 1.9 pawns
E0.5P    - count of positions where player has blundered by 0.5 to 0.9 pawns
EECnt    - count of positions where player move score is better than engine move score
EERate   - 100*EECnt/TotalPos or the percentage of the Engine Error Counts

Notes:
1. Analysis of positions start at move 12.
2. If best score of engine is already above 3 pawns and the score of player move
   is also above 3 pawns then error for this position is set to zero. The more cases
   like this only proves that that player is able to maintain its advantage and this
   will minimize its average error.
   On the other hand if the best score of engine is already below minus 3 pawns, no further
   analysis will be done, and this position will not be included in error calculation.
3. When a player move is a blunder that brings down the score below minus 3 pawns,
   error for this position is set to 6 pawns in the calculation. The reason being once
   the opponent found the refutation, the succeeding positions will no longer be evaluated.
   It is just fair to penalize this heavily to also reflect on the average error which
   should be high for players that have more game loses or fatal blunders.
4. move_error = engine_score - player_score.
   engine_score = score returned by the engine after searching for a given amount of time.
   player_score = score returned by the engine after searching the player move for a given amount of time.
   AveError = sum of [move_error] / totalPos.
5. Error is increased if the score of position after player move has dropped to below zero.
   6x error if (score <= -200 cp), 5x error if (score <= -150 cp), 4x error if (score <= -100 cp),
   3x error if (score <= -50 cp), and 2x error if (score < 0 cp).
   The idea is only those errors which will result to inferior position (below zero score) should be penalized more.
   If the increased error penalty is more than 3 pawns, error is set to 3 pawns.
6. Regarding EECnt or Engine error counts, this is a counter check on engine where the human move
   is sometimes better than the engine move, this is because the analysis time given to the engine is
   shorter and also perhaps the position is complicated that bigger depths are required to determine
   correct assessment of the position.
Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

Comparison New York 1927.

Code: Select all

Move Error Evaluator v8.5

Moves Error calculation based from Stockfish 180415 64 POPCNT and players from
Event New York 1927, after round 1

Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Engine id name Stockfish 180415 64 POPCNT
Hash 128 MB, Threads 1, MoveTime 1.0 sec/pos

 Rank                       Players  AveDepth  TotalPos  AveError  E3.0P  E2.0P  E1.0P  E0.5P  EECnt   EERate    W    L    D  Games  Score  SRate

    1         Capablanca, Jose Raul        18       147      6.54      0      0      0      5     11    7.48%    3    0    2      5    4.0   80.0%
    2            Nimzowitsch, Aaron        20       148     11.59      0      0      1      3     11    7.43%    3    1    1      5    3.5   70.0%
    3           Alekhine, Alexander        18       166     25.31      0      0      2     11      9    5.42%    1    1    3      5    2.5   50.0%
    4             Spielmann, Rudolf        19       155     35.63      0      0      3      9      8    5.16%    0    2    3      5    1.5   30.0%
    5         Marshall, Frank James        19       145     44.25      1      0      3      7     11    7.59%    0    2    3      5    1.5   30.0%
    6              Vidmar, Milan Sr        18       119     44.32      0      1      2      8      6    5.04%    0    1    4      5    2.0   40.0%

Overall Average Error: 27.94
Astatos
Posts: 18
Joined: Thu Apr 10, 2014 5:20 pm

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Astatos »

So, in other words, you proved that your method(s) prove(s) nothing :D
Vinvin
Posts: 5308
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Vinvin »

Ferdy wrote:

Code: Select all

Err_3.0P - count of positions where player has blundered by 3 or more pawns
Err_2.0P - count of positions where player has blundered by 2 to 2.9 pawns
Err_1.0P - count of positions where player has blundered by 1 to 1.9 pawns
Err_0.5P - count of positions where player has blundered by 0.5 to 0.9 pawns
Please add blunder by 0.2 to 0.49 !
Ferdy
Posts: 4851
Joined: Sun Aug 10, 2008 3:15 pm
Location: Philippines

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Ferdy »

Vinvin wrote:
Ferdy wrote:

Code: Select all

Err_3.0P - count of positions where player has blundered by 3 or more pawns
Err_2.0P - count of positions where player has blundered by 2 to 2.9 pawns
Err_1.0P - count of positions where player has blundered by 1 to 1.9 pawns
Err_0.5P - count of positions where player has blundered by 0.5 to 0.9 pawns
Please add blunder by 0.2 to 0.49 !
Here it is, also with new info on dangerous error counts, DECnt. This is the count of positions where players' position goes below zero after a move. This feature will give us info how many times a player is in trouble.

Sample table sorted on less DECnt, from sr5_less_fatal_error.txt.

Code: Select all

Players                   AveDepth  TotalPos  AveError  DECnt   E300CP  E200CP  E100CP  E50CP   E20CP   EECnt   EERate  W    L    D    Games  Score  SRate
------------------------  --------  --------  --------  ------  ------  ------  ------  ------  ------  ------  ------  ---  ---  ---  -----  -----  -----
   Carlsen, Magnus        13        288       9.54      25      0       0       1       5       26      29      10.07   5    0    4    9      7      77.8 
        So, Wesley        12        284       15.2      42      0       0       2       10      25      29      10.21   3    2    4    9      5      55.6 
Anand, Viswanathan        12        280       12.76     47      0       0       3       8       32      28      10.0    3    0    6    9      6      66.7 
     Mamedov, Rauf        12        262       16.29     48      0       0       2       6       32      27      10.31   0    2    7    9      3.5    38.9 
       Giri, Anish        12        299       19.96     51      0       0       6       11      28      31      10.37   0    2    7    9      3.5    38.9 
Mamedyarov, Shakhriyar    12        307       14.65     52      1       0       2       8       32      31      10.1    1    2    6    9      4      44.4 
Vachier-Lagrave, Maxime   13        310       17.3      59      0       1       2       8       22      34      10.97   0    2    7    9      3.5    38.9 
  Caruana, Fabiano        13        327       19.09     59      0       0       3       13      37      30      9.17    2    1    6    9      5      55.6 
 Kramnik, Vladimir        12        337       26.97     71      0       0       3       14      33      32      9.5     2    3    4    9      4      44.4 
    Adams, Michael        12        258       25.31     73      1       0       1       10      36      22      8.53    1    3    5    9      3.5    38.9 
Then we have the ave error table, from sr4_less_ave_error.txt.

Code: Select all

Players                   AveDep  TotalPos  AveError  DECnt   E300CP  E200CP  E100CP  E50CP   E20CP   EECnt   EERate  W    L    D    Games  Score  SRate
------------------------  ------  --------  --------  ------  ------  ------  ------  ------  ------  ------  ------  ---  ---  ---  -----  -----  -----
   Carlsen, Magnus        13      288       9.54      25      0       0       1       5       26      29      10.07   5    0    4    9      7      77.8 
Anand, Viswanathan        12      280       12.76     47      0       0       3       8       32      28      10.0    3    0    6    9      6      66.7 
Mamedyarov, Shakhriyar    12      307       14.65     52      1       0       2       8       32      31      10.1    1    2    6    9      4      44.4 
        So, Wesley        12      284       15.2      42      0       0       2       10      25      29      10.21   3    2    4    9      5      55.6 
     Mamedov, Rauf        12      262       16.29     48      0       0       2       6       32      27      10.31   0    2    7    9      3.5    38.9 
Vachier-Lagrave, Maxime   13      310       17.3      59      0       1       2       8       22      34      10.97   0    2    7    9      3.5    38.9 
  Caruana, Fabiano        13      327       19.09     59      0       0       3       13      37      30      9.17    2    1    6    9      5      55.6 
       Giri, Anish        12      299       19.96     51      0       0       6       11      28      31      10.37   0    2    7    9      3.5    38.9 
    Adams, Michael        12      258       25.31     73      1       0       1       10      36      22      8.53    1    3    5    9      3.5    38.9 
 Kramnik, Vladimir        12      337       26.97     71      0       0       3       14      33      32      9.5     2    3    4    9      4      44.4 
This is just a quick sample using sf6 at 10 ms / position. As you can see the EECnt (engine error count) is high and the ave depth is low.
I will give more analysis time later.

Standings, sorted by score and num wins. Last 3 ranking is not accurate. Nice output from sqlite cli.

Code: Select all

Players                   AveDepth  TotalPos  AveError  DECnt   E300CP  E200CP  E100CP  E50CP   E20CP   EECnt   EERate  W    L    D    Games  Score  SRate
------------------------  --------  --------  --------  ------  ------  ------  ------  ------  ------  ------  ------  ---  ---  ---  -----  -----  -----
   Carlsen, Magnus        13        288       9.54      25      0       0       1       5       26      29      10.07   5    0    4    9      7      77.8 
Anand, Viswanathan        12        280       12.76     47      0       0       3       8       32      28      10.0    3    0    6    9      6      66.7 
        So, Wesley        12        284       15.2      42      0       0       2       10      25      29      10.21   3    2    4    9      5      55.6 
  Caruana, Fabiano        13        327       19.09     59      0       0       3       13      37      30      9.17    2    1    6    9      5      55.6 
 Kramnik, Vladimir        12        337       26.97     71      0       0       3       14      33      32      9.5     2    3    4    9      4      44.4 
Mamedyarov, Shakhriyar    12        307       14.65     52      1       0       2       8       32      31      10.1    1    2    6    9      4      44.4 
    Adams, Michael        12        258       25.31     73      1       0       1       10      36      22      8.53    1    3    5    9      3.5    38.9 
     Mamedov, Rauf        12        262       16.29     48      0       0       2       6       32      27      10.31   0    2    7    9      3.5    38.9 
Vachier-Lagrave, Maxime   13        310       17.3      59      0       1       2       8       22      34      10.97   0    2    7    9      3.5    38.9 
       Giri, Anish        12        299       19.96     51      0       0       6       11      28      31      10.37   0    2    7    9      3.5    38.9 
Guenther
Posts: 4718
Joined: Wed Oct 01, 2008 6:33 am
Location: Regensburg, Germany
Full name: Guenther Simon

Re: SF6 moves comparison: Event Vugar Gashimov Mem 2015

Post by Guenther »

Hi Ferdinand,

You should also exclude one time repetition moves, because Humans just play them to gain time to reach the time control and have more time to calculate the winning line.
If the weaker side really has nothing better then to repeat too, there is nothing lost for the better side!

Guenther