Dev Builds » 20240320-1529

Use this dev build

NCM plays each Stockfish dev build 20,000 times against Stockfish 15. This yields an approximate Elo difference and establishes confidence in the strength of the dev builds.

Summary

Host Duration Avg Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo
ncm-dbt-01 06:52:28 584907 4000 1516 603 1881 +80.72 ± 4.91 1 112 870 1007 10 +169.27 ± 11.5
ncm-dbt-02 06:52:14 584846 4000 1473 566 1961 +80.17 ± 4.89 2 108 880 1001 9 +168.4 ± 11.42
ncm-dbt-03 06:50:50 585196 4000 1488 551 1961 +82.93 ± 4.89 0 113 844 1036 7 +174.99 ± 11.69
ncm-dbt-04 06:51:50 572818 4000 1462 581 1957 +77.8 ± 4.9 0 119 890 982 9 +162.35 ± 11.36
ncm-dbt-05 06:52:47 586945 4000 1458 629 1913 +73.06 ± 5.02 2 135 910 938 15 +150.51 ± 11.25
20000 7397 2930 9673 +78.93 ± 2.2 5 587 4394 4964 50 +165.02 ± 5.12

Test Detail

ID Host Base NPS Games WLD Standard Elo Ptnml(0-2) Gamepair Elo CLI PGN
380712 ncm-dbt-05 589539 500 184 81 235 +72.61 ± 13.39 0 14 119 117 0 +152.18 ± 31.0 ↓
380711 ncm-dbt-01 585000 500 189 70 241 +84.31 ± 14.47 0 19 93 138 0 +179.9 ± 35.49 ↓
380710 ncm-dbt-02 590653 500 187 71 242 +82.1 ± 14.56 1 16 100 132 1 +174.55 ± 34.22 ↓
380709 ncm-dbt-04 573608 500 180 85 235 +66.82 ± 13.63 0 16 124 109 1 +137.37 ± 30.31 ↓
380708 ncm-dbt-03 586604 500 191 71 238 +85.04 ± 14.35 0 17 97 135 1 +179.9 ± 34.76 ↓
380707 ncm-dbt-05 584622 500 174 75 251 +69.71 ± 14.59 0 21 111 116 2 +142.26 ± 32.42 ↓
380706 ncm-dbt-01 586139 500 194 71 235 +87.26 ± 14.1 1 11 104 132 2 +185.33 ± 33.42 ↓
380705 ncm-dbt-02 583363 500 194 66 240 +90.96 ± 13.08 0 7 110 131 2 +192.71 ± 32.1 ↓
380704 ncm-dbt-04 569549 500 177 72 251 +74.06 ± 14.46 0 20 106 123 1 +153.86 ± 33.21 ↓
380703 ncm-dbt-03 584958 500 184 70 246 +80.63 ± 13.94 0 14 110 124 2 +167.53 ± 32.45 ↓
380702 ncm-dbt-05 586350 500 185 93 222 +64.66 ± 13.69 0 17 125 107 1 +132.54 ± 30.2 ↓
380701 ncm-dbt-01 583950 500 185 84 231 +71.16 ± 13.64 0 15 120 114 1 +147.19 ± 30.88 ↓
380700 ncm-dbt-02 583866 500 190 75 235 +81.36 ± 14.25 0 15 108 124 3 +167.53 ± 32.81 ↓
380699 ncm-dbt-04 572437 500 194 66 240 +90.96 ± 13.87 0 13 97 139 1 +194.57 ± 34.74 ↓
380698 ncm-dbt-03 584579 500 182 79 239 +72.61 ± 13.84 0 16 116 117 1 +150.51 ± 31.54 ↓
380697 ncm-dbt-05 589497 500 189 80 231 +76.98 ± 14.13 1 13 114 120 2 +160.64 ± 31.79 ↓
380696 ncm-dbt-01 584243 500 199 72 229 +90.22 ± 13.39 0 10 104 135 1 +192.71 ± 33.34 ↓
380695 ncm-dbt-02 582903 500 183 66 251 +82.83 ± 13.54 0 13 107 130 0 +176.33 ± 32.93 ↓
380694 ncm-dbt-04 572598 500 181 74 245 +75.52 ± 13.64 0 15 113 122 0 +158.93 ± 31.99 ↓
380693 ncm-dbt-03 584664 500 184 72 244 +79.17 ± 12.98 0 10 118 122 0 +167.53 ± 30.96 ↓
380692 ncm-dbt-05 587579 500 186 71 243 +81.36 ± 14.68 0 17 105 124 4 +165.8 ± 33.35 ↓
380691 ncm-dbt-01 584874 500 180 81 239 +69.71 ± 13.74 0 17 117 116 0 +145.54 ± 31.41 ↓
380690 ncm-dbt-02 581902 500 181 73 246 +76.25 ± 13.05 0 10 123 116 1 +158.93 ± 30.17 ↓
380689 ncm-dbt-04 574906 500 185 65 250 +85.04 ± 13.28 0 10 111 128 1 +179.9 ± 32.11 ↓
380688 ncm-dbt-03 586308 500 180 68 252 +79.17 ± 13.45 0 13 112 125 0 +167.53 ± 32.08 ↓
380687 ncm-dbt-05 586097 500 180 67 253 +79.9 ± 13.92 0 15 108 126 1 +167.53 ± 32.81 ↓
380686 ncm-dbt-01 583447 500 195 82 223 +79.9 ± 13.47 0 10 120 117 3 +164.07 ± 30.64 ↓
380685 ncm-dbt-03 583950 500 187 62 251 +88.74 ± 14.57 0 17 93 138 2 +187.16 ± 35.51 ↓
380684 ncm-dbt-02 585169 500 187 76 237 +78.44 ± 13.88 1 13 110 126 0 +167.53 ± 32.45 ↓
380683 ncm-dbt-04 572437 500 185 69 246 +82.1 ± 13.83 0 12 113 122 3 +169.27 ± 31.88 ↓
380682 ncm-dbt-01 586308 500 181 73 246 +76.25 ± 14.1 0 18 106 126 0 +160.64 ± 33.2 ↓
380681 ncm-dbt-03 585632 500 198 57 245 +100.7 ± 12.52 0 5 99 146 0 +221.9 ± 33.98 ↓
380680 ncm-dbt-05 588558 500 183 82 235 +71.16 ± 14.08 0 17 117 114 2 +145.54 ± 31.41 ↓
380679 ncm-dbt-02 588302 500 184 59 257 +88.74 ± 13.98 0 13 101 134 2 +187.16 ± 33.99 ↓
380678 ncm-dbt-04 574337 500 188 80 232 +76.25 ± 14.1 0 17 109 123 1 +158.93 ± 32.69 ↓
380677 ncm-dbt-03 584874 500 182 72 246 +77.7 ± 14.72 0 21 99 129 1 +162.35 ± 34.39 ↓
380676 ncm-dbt-05 583321 500 177 80 243 +68.27 ± 15.08 1 21 111 114 3 +138.99 ± 32.43 ↓
380675 ncm-dbt-02 582611 500 167 80 253 +61.07 ± 13.96 0 21 121 108 0 +126.18 ± 30.9 ↓
380674 ncm-dbt-01 585295 500 193 70 237 +87.26 ± 13.95 0 12 106 129 3 +181.7 ± 33.07 ↓
380673 ncm-dbt-04 572679 500 172 70 258 +71.88 ± 13.81 0 16 117 116 1 +148.85 ± 31.38 ↓

Commit

Commit ID 9b92ada935ddf920491156be22f609afaca4d840
Author Robert Nurnberg
Date 2024-03-20 15:29:35 UTC
Base WDL model on material count and normalize evals dynamically This PR proposes to change the parameter dependence of Stockfish's internal WDL model from full move counter to material count. In addition it ensures that an evaluation of 100 centipawns always corresponds to a 50% win probability at fishtest LTC, whereas for master this holds only at move number 32. See also https://github.com/official-stockfish/Stockfish/pull/4920 and the discussion therein. The new model was fitted based on about 340M positions extracted from 5.6M fishtest LTC games from the last three weeks, involving SF versions from e67cc979fd2c0e66dfc2b2f2daa0117458cfc462 (SF 16.1) to current master. The involved commands are for [WDL_model](https://github.com/official-stockfish/WDL_model) are: ``` ./updateWDL.sh --firstrev e67cc979fd2c0e66dfc2b2f2daa0117458cfc462 python scoreWDL.py updateWDL.json --plot save --pgnName update_material.png --momType "material" --momTarget 58 --materialMin 10 --modelFitting optimizeProbability ``` The anchor `58` for the material count value was chosen to be as close as possible to the observed average material count of fishtest LTC games at move 32 (`43`), while not changing the value of `NormalizeToPawnValue` compared to the move-based WDL model by more than 1. The patch only affects the displayed cp and wdl values. closes https://github.com/official-stockfish/Stockfish/pull/5121 No functional change
Copyright 2011–2024 Next Chess Move LLC
Feedback & Support Terms of Service Privacy Policy Desktop Version