Stockfish Testing Queue

Pending - 0 tests 0.0 hrs

None

Active - 0 tests

Finished - 2076 tests

19-06-14 Roc PawnPushBlocker2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 10033 W: 2163 L: 2278 D: 5592
sprt @ 10+0.1 th 1 PPB2_2 Take 2: a more precise version. Allow pawn push though our own minors and rook as long as unpinned. exclude pawn push on squares controlled by enemy (so we exclude the en paant case)
19-06-14 Roc PawnPushBlocker diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 11001 W: 2391 L: 2501 D: 6109
sprt @ 10+0.1 th 1 PPB_1 Back to simpler stuff. White f2 Nf3, black Ne4 Since Nf3 can move away, consider that Ne4 is threatened by a pawn push.
19-06-14 Roc PawnPushAndOutposts2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 6313 W: 1409 L: 1544 D: 3360
sprt @ 10+0.1 th 1 Fixed another error in my previous test.
19-06-14 Roc PawnPushAndOutposts2 diff
LLR: -1.97 (-2.94,2.94) [0.50,4.50]
Total: 1504 W: 306 L: 410 D: 788
sprt @ 10+0.1 th 1 Previous 2 tests had a huge flaw.This is now corrected, Let's see if we have some progress, before excluding more cases. Run as normal test
19-06-14 Roc PawnPushAndOutposts diff
LLR: -1.85 (-2.94,2.94) [-3.00,1.00]
Total: 3258 W: 699 L: 810 D: 1749
sprt @ 10+0.1 th 1 Take 2
19-06-14 Roc PawnPushAndOutposts diff
LLR: -1.17 (-2.94,2.94) [-3.00,1.00]
Total: 1645 W: 337 L: 406 D: 902
sprt @ 10+0.1 th 1 outposts can't be reached by a threatbypawnpush. Try to unify logic and hope for an improvement.
19-06-14 Roc SimpleOutpost2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 19400 W: 4273 L: 4341 D: 10786
sprt @ 10+0.1 th 1 Retry an old idea in the new context of pawn protected outposts
19-06-13 Roc KoTuned2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 34201 W: 7642 L: 7636 D: 18923
sprt @ 10+0.1 th 1 KT2 Using outpost model by rank, including rank_3 Using tuned values after 100M games at STC
19-06-12 Roc KOTuning2 diff
46245/50000 iterations
100000/100000 games played
100000 @ 20+0.2 th 1 Fixed bench. Tuning minor outpost and psqt ranks (including rank 3)
19-06-10 Roc TunedKnightPQT diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 61545 W: 13865 L: 13784 D: 33896
sprt @ 10+0.1 th 1 TKP_1 Tuned values after 100M games
19-06-10 Roc TuneKnightPQT diff
44768/50000 iterations
99764/100000 games played
100000 @ 20+0.2 th 1 Outpost bonus are now only for pawn protected outposts. Tune Knight PSQT
19-06-09 Roc UnsafeMobility diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 7880 W: 1727 L: 1853 D: 4300
sprt @ 10+0.1 th 1 UM_1 Use master values if we can exchange against another piece otheriwes use a lower mobility value. To make this work simply add a new (lower) score at position [0]
19-06-09 Roc BackwardPhalanx3 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 14330 W: 3133 L: 3227 D: 7970
sprt @ 10+0.1 th 1 Take 2: try a middlegame bonus
19-06-09 Roc BackwardPhalanx3 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 18909 W: 4154 L: 4225 D: 10530
sprt @ 10+0.1 th 1 BP3 Allocate a bonus for a pawn pattern which is seen often in french or english defense. (e.g White e5 h5 against black e6, f7, g7, h6
19-05-25 Roc BrokenChainBackward diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 25799 W: 5766 L: 5802 D: 14231
sprt @ 10+0.1 th 1 BCB_2 Take 2, 50% more
19-05-25 Roc BrokenChainBackward diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 11102 W: 2485 L: 2595 D: 6022
sprt @ 10+0.1 th 1 Bkw More backward penalty if disconnected from some neighbour
19-05-25 Roc ProtectedOutpost diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 23051 W: 5099 L: 5149 D: 12803
sprt @ 10+0.1 th 1 PO_1 White b5 d5 Rc6 Nc3 Black Bb7 Although b5 and d5 are currently occupied by pawns, the ability to jump on those squares if the rook is captured is a positional asset. Take 1: full reachable bonus for b5 or d5, assuming it defends a minor or major outpost on c6
19-05-25 Roc BrokenChainBackward diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 14381 W: 3172 L: 3265 D: 7944
sprt @ 10+0.1 th 1 BCB_3 more penatty only if 1) 2 neighbours 2) and one of them not supported by the backward
19-05-20 Roc ExposedPaths4 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 9864 W: 2169 L: 2285 D: 5410
sprt @ 10+0.1 th 1 EP4 Take 3 was considering only knights, and was worse then expected. Try only bishops and rooks.
19-05-19 Roc ExposedPaths3 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 11846 W: 2609 L: 2715 D: 6522
sprt @ 10+0.1 th 1 EP3 Take 3, simple version, bonus for knights which can attack unsupported pawns on next move
19-05-19 Roc ExposedPaths2 diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 11526 W: 2555 L: 2663 D: 6308
sprt @ 10+0.1 th 1 EP2 Take 2: exclude phalanx pawns
19-05-19 Roc ExposedPaths diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12156 W: 2690 L: 2795 D: 6671
sprt @ 10+0.1 th 1 Calculate different angles of attack to unsupported pawns in pawns.cpp: knight,bishop45, bishop135, rank and file In evaluate.cpp, find "safe" one movers which would add a new attack.
19-05-18 Roc MaxROP diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 17070 W: 3780 L: 3860 D: 9430
sprt @ 10+0.1 th 1 MR_4 Take 2; at most factor 4
19-05-18 Roc WeakPawnSpan diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 9648 W: 2074 L: 2191 D: 5383
sprt @ 10+0.1 th 1 Fixed bench typo
19-05-18 Roc MaxROP diff
LLR: -2.94 (-2.94,2.94) [0.50,4.50]
Total: 9516 W: 2041 L: 2158 D: 5317
sprt @ 10+0.1 th 1 MR_3 Cap the RookOnPawn factor to 3
19-05-03 Roc firstRankminors diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 17085 W: 3718 L: 3798 D: 9569
sprt @ 10+0.1 th 1 smaller value
19-05-03 Roc firstRankminors diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 21033 W: 4616 L: 4676 D: 11741
sprt @ 10+0.1 th 1 frm-q6
19-05-02 Roc firstRankminors diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 29931 W: 6676 L: 6692 D: 16563
sprt @ 10+0.1 th 1 frM-Q Penalize adventurous queen when minors on first rank
19-05-02 Roc firstRankminors diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 7771 W: 1656 L: 1782 D: 4333
sprt @ 10+0.1 th 1 Double the penalty
19-05-02 Roc firstRankminors diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 9940 W: 2157 L: 2273 D: 5510
sprt @ 10+0.1 th 1 frm20
19-05-02 Roc firstRankminors diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 4864 W: 1038 L: 1179 D: 2647
sprt @ 10+0.1 th 1 frm40
19-04-30 Roc LowMobInteraction diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12930 W: 2817 L: 2918 D: 7195
sprt @ 10+0.1 th 1 LMI-2
19-04-20 Roc BD_15 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 28297 W: 6327 L: 6379 D: 15591
sprt @ 10+0.1 th 1 Eg value for bishop diagonal try 15
19-04-19 Roc Pawns8mTuned diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 38175 W: 7140 L: 7164 D: 23871
sprt @ 10+0.1 th 1 Run again as SPRT[0, 4] with 8 moves v3
19-04-19 Roc Pawns8mTuned diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 19028 W: 4131 L: 4220 D: 10677
sprt @ 10+0.1 th 1 Run again as SPRT[0, 4]
19-04-18 Roc Pawns8mTuned diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 15582 W: 3433 L: 3520 D: 8629
sprt @ 10+0.1 th 1 P*m-2 after 47135 iterations
19-04-18 Roc PTune8m diff
47135/50000 iterations
99369/100000 games played
100000 @ 20+0.2 th 1 Try pawn tuning using 8 moves book with book depth = 6
19-04-18 Roc TwoMinors diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 6431 W: 1355 L: 1488 D: 3588
sprt @ 10+0.1 th 1 TM_10_20 Penalty for patterns such as Ra1 Nb1 Bc1 (lack of development) or Re1 Ne4 Be5 (minorsin the way)
19-04-18 Roc Zugz1 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 13559 W: 2988 L: 3085 D: 7486
sprt @ 10+0.1 th 1 Z1_RBK_4 rook + bishop + king, S(0, 4) per square
19-04-18 Roc MinkpDistanceCap diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 20040 W: 4404 L: 4469 D: 11167
sprt @ 10+0.1 th 1 MKP4 Fixed version.
19-04-18 Roc MinkpDistanceCap diff
LLR: -0.11 (-2.94,2.94) [0.50,4.50]
Total: 233 W: 41 L: 46 D: 146
sprt @ 10+0.1 th 1 do not overscore this penalty, apply only for distance < 4
19-04-14 Roc PassedFile2 diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 31292 W: 6945 L: 6954 D: 17393
sprt @ 10+0.1 th 1 pf2 don,t use passed file for rank 2 or 3
19-04-14 Roc PP_k diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 8605 W: 1831 L: 1953 D: 4821
sprt @ 10+0.1 th 1 fixed bench, not my day
19-04-14 Roc PP_k diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 12303 W: 2690 L: 2794 D: 6819
sprt @ 10+0.1 th 1 Since passedFile is a large penalty for center files apply it only for rank > 4 to avoid scoring some passed pawns negatively. While being there, try to simplify a bit. Fixed.
19-04-14 Roc PP_k diff
LLR: -1.13 (-2.94,2.94) [0.50,4.50]
Total: 108 W: 7 L: 75 D: 26
sprt @ 10+0.1 th 1 PP_k1 Since passedFile is a large penalty for center files apply it only for rank > 4 to avoid scoring some passed pawns negatively. While being there, try to simplify a bit.
19-04-13 Roc SimplePawnConn diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 18670 W: 4173 L: 4245 D: 10252
sprt @ 10+0.1 th 1 spc_5 With factor such as 2 3, 4 or 6, each change is quite sensitive Try again to get rid of negative values by decrease also mg.
19-04-13 Roc SimplePawnConn diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 12857 W: 2837 L: 2938 D: 7082
sprt @ 10+0.1 th 1 SPC_4 Try tuned values without the negative values
19-04-13 Roc SimplePawnConn diff
LLR: -2.96 (-2.94,2.94) [0.50,4.50]
Total: 54459 W: 12277 L: 12170 D: 30012
sprt @ 10+0.1 th 1 SPC_3 Fixed error in rewrite
19-04-13 Roc SimplePawnConn diff
LLR: -2.95 (-2.94,2.94) [0.50,4.50]
Total: 4216 W: 894 L: 1039 D: 2283
sprt @ 10+0.1 th 1 SPC_2 using tuned result from http://tests.stockfishchess.org/tests/view/5cb0f45a0ebc5925cf013a28
19-04-12 Roc TuneConnMgEg diff
38286/40000 iterations
79811/80000 games played
80000 @ 20+0.2 th 1 Tune new connected.While being there, try more accurate eg values for connected. Fixed bench