Stockfish Testing Queue

Finished - 19467 tests

13-11-13 gl master diff
ELO: 0.15 +-1.9 (95%) LOS: 56.1%
Total: 40000 W: 6136 L: 6119 D: 27745
40000 @ 60+0.05 th 1 Long TC: Aggressiveness=95 Cowardice=95
15-11-13 gl safety_weight diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 24184 W: 4525 L: 4567 D: 15092
sprt @ 15+0.05 th 1 SPRT: Increase internal king danger weight (us).
15-11-13 gl null_depth diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 12175 W: 2300 L: 2373 D: 7502
sprt @ 15+0.05 th 1 Null depth (8 * depth + 164) / 32
15-11-13 gl null_depth diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 24611 W: 4642 L: 4682 D: 15287
sprt @ 15+0.05 th 1 Null depth (7 * depth + 164) / 32
15-11-13 gl null_depth diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 38021 W: 7249 L: 7253 D: 23519
sprt @ 15+0.05 th 1 Null depth (7 * depth + 184) / 32
15-11-13 ee approximateEval diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 17683 W: 3302 L: 3361 D: 11020
sprt @ 15+0.05 th 1 After a null move we can skip the evaluation. With new search from Lucas evalMargin (King Safety asymmetry) should be less of a problem and is no longer computed. In my slightly different local version the bench is lower but for this version unfortunately not.
15-11-13 mc null_depth diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 5231 W: 805 L: 883 D: 3543
sprt @ 60+0.05 th 1 LTC: Null depth (9 * depth + 164) / 32
14-11-13 gl master diff
ELO: 0.43 +-2.1 (95%) LOS: 65.5%
Total: 33377 W: 5272 L: 5231 D: 22874
40000 @ 60+0.05 th 1 Long TC: Aggressiveness=105 Cowardice=10
15-11-13 gl null_depth diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 17520 W: 3426 L: 3273 D: 10821
sprt @ 15+0.05 th 1 Null depth (9 * depth + 164) / 32
15-11-13 Fi hash_static_eval diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 11599 W: 2173 L: 2248 D: 7178
sprt @ 15+0.05 th 1 Store static eval values in the TT after they are computed.
14-11-13 rs easy diff
LLR: -1.42 (-2.94,2.94) [-1.50,4.50]
Total: 6286 W: 1152 L: 1186 D: 3948
sprt @ 15+0.05 th 1 easy move experiment while framework is empty
14-11-13 gl master diff
ELO: -2.50 +-2.8 (95%) LOS: 3.8%
Total: 18656 W: 2803 L: 2937 D: 12916
40000 @ 60+0.05 th 1 Long TC: Aggressiveness=110 Cowardice=105
14-11-13 rs piece_bonuses diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 8288 W: 1487 L: 1570 D: 5231
sprt @ 15+0.05 th 1 award piece bonuses only if the pieces are not attacked by a pawn (same as BishopPin bonus)
14-11-13 lb clop diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 13230 W: 2475 L: 2546 D: 8209
sprt @ 15+0.05 th 1 CLOP futility margins
13-11-13 gl master diff
ELO: 1.05 +-1.9 (95%) LOS: 86.1%
Total: 40000 W: 6264 L: 6143 D: 27593
40000 @ 60+0.05 th 1 Long TC: Aggressiveness=105 Cowardice=105
12-11-13 lb master diff
ELO: 63.85 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 11454 L: 4185 D: 24361
40000 @ 60+0.05 th 1 Regression test with "Reintroduce gains" and "Simplify evasions"
13-11-13 rs ks_no_checks diff
ELO: -16.38 +-3.8 (95%) LOS: 0.0%
Total: 12992 W: 2286 L: 2898 D: 7808
20000 @ 15+0.05 th 1 need some info about the value of check bonuses in king safety function
14-11-13 lb null_red diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 2020 W: 330 L: 430 D: 1260
sprt @ 15+0.05 th 1 tweak null reduction. Try using DiscoCheck's CLOP'ed formula.
13-11-13 rs pp_tune diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 18874 W: 3530 L: 3586 D: 11758
sprt @ 15+0.05 th 1 try to simplify and tune supportedPawns bonus (take 2 bugfix)
13-11-13 gl null_depth diff
ELO: -7.16 +-3.6 (95%) LOS: 0.0%
Total: 10870 W: 1580 L: 1804 D: 7486
40000 @ 60+0.05 th 1 16*d+114 (more aggressive)
13-11-13 rs pp_tune diff
LLR: -2.53 (-2.94,2.94) [-1.50,4.50]
Total: 6833 W: 1217 L: 1289 D: 4327
sprt @ 15+0.05 th 1 try to simplify and tune supportedPawns bonus
13-11-13 gl null_depth diff
ELO: -8.49 +-4.4 (95%) LOS: 0.0%
Total: 7287 W: 1019 L: 1197 D: 5071
40000 @ 60+0.05 th 1 Try Lucas' null-move clop values
13-11-13 gl master diff
ELO: -1.98 +-2.1 (95%) LOS: 2.9%
Total: 40000 W: 7174 L: 7402 D: 25424
40000 @ 15+0.05 th 1 Aggressiveness=95 Cowardice=105
11-11-13 gl master diff
ELO: 61.23 +-2.1 (95%) LOS: 100.0%
Total: 40000 W: 11310 L: 4333 D: 24357
40000 @ 60+0.05 th 1 Regression test after material simplfication + eval_pin bugfix
13-11-13 rs ks_tune diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 15569 W: 2833 L: 2898 D: 9838
sprt @ 15+0.05 th 1 KnightCheck=BishopCheck=3
13-11-13 gl master diff
ELO: -0.26 +-2.1 (95%) LOS: 40.4%
Total: 40000 W: 7530 L: 7560 D: 24910
40000 @ 15+0.05 th 1 Aggressiveness=105 Cowardice=95
13-11-13 rs ks_tune diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 4213 W: 773 L: 868 D: 2572
sprt @ 15+0.05 th 1 try a different set of distance and contact check params
13-11-13 gl master diff
ELO: -0.45 +-2.1 (95%) LOS: 33.6%
Total: 40000 W: 7506 L: 7558 D: 24936
40000 @ 15+0.05 th 1 Aggressiveness=105 Cowardice=105
13-11-13 rs ks_open diff
LLR: -0.94 (-2.94,2.94) [-1.50,4.50]
Total: 7748 W: 1452 L: 1465 D: 4831
sprt @ 15+0.05 th 1 try a king safety tweak regarding empty squares around the king
13-11-13 gl master diff
ELO: -0.30 +-2.1 (95%) LOS: 39.0%
Total: 40000 W: 7428 L: 7462 D: 25110
40000 @ 15+0.05 th 1 Aggressiveness=95 Cowardice=95
06-10-13 gl master diff
Status: 17545 3025 3403 0.470597 84 124
Total: 23565
64000 @ 15+0.05 th 1 CLOP test take 2
12-11-13 ce opp_b diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 51266 W: 9568 L: 9539 D: 32159
sprt @ 15+0.05 th 1 Reduce approx 20% except for passed pawns
12-11-13 rs ignore_safe diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 278 W: 36 L: 159 D: 83
sprt @ 15+0.05 th 1 In king safety we award very high bonuses for enemy pieces that threaten contact checks, even if it's our turn and we could capture those pieces. To ignore tactical ad-hoc considerations seems to work well in eval, therefore I want to try to also ignore the "safe" condition for distance checks. It is a tiny simplification, Let's see how it goes since framework is nearly empty.
12-11-13 mc eval diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 15424 W: 2888 L: 2953 D: 9583
sprt @ 15+0.05 th 1 Retire bishop pin bonus, now that we already penalize pinned pieces.
12-11-13 ce opp_b^ diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 3322 W: 546 L: 642 D: 2134
sprt @ 15+0.05 th 1 Reduce approx 30% except passed pawns
12-11-13 rs qcc_prepare2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 25461 W: 4771 L: 4809 D: 15881
sprt @ 15+0.05 th 1 Queen contact check preparation. Final attempt to improve on take 1 that failed at LTC with positive score.
12-11-13 gl master diff
Status: -
Total: 188
64000 @ 15+0.05 th 1 CLOP test with Mobility and Pawn Structure.
12-11-13 ce opp_b diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 7568 W: 1342 L: 1427 D: 4799
sprt @ 15+0.05 th 1 Half opp. bishop scoring except for passed pawns
12-11-13 ce mate_threat diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 8839 W: 1652 L: 1734 D: 5453
sprt @ 15+0.05 th 1 Bigger bonus for mate threat. Last try while queue short.
11-11-13 re remove_minorbp diff
ELO: -3.94 +-2.9 (95%) LOS: 0.4%
Total: 20000 W: 3594 L: 3821 D: 12585
20000 @ 15+0.05 th 1 Evaluate minor behind pawn bonus.
12-11-13 rs qcc_prepare2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 9564 W: 1734 L: 1814 D: 6016
sprt @ 15+0.05 th 1 Consider preparation of queen contact checks (final take 4)
12-11-13 jh queen_contact_pred diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5621 W: 1017 L: 1108 D: 3496
sprt @ 15+0.05 th 1 Variation of Ralph Stoesser's queen contact idea.
11-11-13 rs qcc_prepare1 diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 15553 W: 2892 L: 2956 D: 9705
sprt @ 15+0.05 th 1 Consider preparation of queen contact checks (take 3)
10-11-13 jk mc diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 55670 W: 10405 L: 10364 D: 34901
sprt @ 15+0.05 th 1 Reintroduce moveCount in futility pruning
11-11-13 rs qcc_prepare1 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 14183 W: 2625 L: 2693 D: 8865
sprt @ 15+0.05 th 1 Consider preparation of queen contact checks (take 2)
11-11-13 gl master diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 15772 W: 2781 L: 2845 D: 10146
sprt @ 15+0.05 th 1 Test CLOP tuned values after 17.5 games
11-11-13 rs qcc_prepare1 diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 39752 W: 6268 L: 6188 D: 27296
sprt @ 60+0.05 th 1 LTC: Consider preparation of queen contact checks
11-11-13 ce pawns diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 20162 W: 3748 L: 3800 D: 12614
sprt @ 15+0.05 th 1 Pawn imbalance change 2
11-11-13 ce pawns^ diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4540 W: 792 L: 885 D: 2863
sprt @ 15+0.05 th 1 Pawn imbalance change 1
11-11-13 lb futility_margin diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 8768 W: 1634 L: 1716 D: 5418
sprt @ 15+0.05 th 1 futility_margin = 50 * depth + 270