Stockfish Testing Queue

Finished - 3976 tests

14-05-06 sni coordination2 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 7326 W: 1238 L: 1324 D: 4764
sprt @ 15+0.05 th 1 Coordination : rooks and queen
14-05-06 sni coordination2 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 3696 W: 604 L: 699 D: 2393
sprt @ 15+0.05 th 1 Coordination : pawns
14-05-06 sni stochastic diff
LLR: -2.23 (-2.94,2.94) [-1.50,4.50]
Total: 47110 W: 8339 L: 8299 D: 30472
sprt @ 15+0.05 th 1 Stochastic mobility, after http://www.dcs.bbk.ac.uk/~mark/download/ply.pdf
14-05-06 sni pinned_pawns3 diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 49528 W: 7370 L: 7255 D: 34903
sprt @ 60+0.05 th 1 King defense malus for pinned pawns
14-05-07 sni stochastic diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 10052 W: 1428 L: 1485 D: 7139
sprt @ 60+0.05 th 1 Change noise interval to +/-16 centipawns. May I test this version at LTC speculatively, as they say in the paper the effect might be depth dependant?
14-05-08 sni stochastic2 diff
LLR: -1.94 (-2.94,2.94) [-3.00,1.00]
Total: 56309 W: 10182 L: 10382 D: 35745
sprt @ 15+0.05 th 1 Simplification : remove space calculation, and compensate it by stochastic mobility (stochastic amplitude=15)
14-05-08 sni stochastic2 diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 17258 W: 3035 L: 3221 D: 11002
sprt @ 15+0.05 th 1 Simplification : remove space calculation, and compensate it by stochastic mobility (take 2, stochastic amplitude=31)
14-05-08 sni forks diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 7462 W: 1240 L: 1325 D: 4897
sprt @ 15+0.05 th 1 Try some tweaks on hanging patch.
14-05-09 sni outpost5 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 6179 W: 1077 L: 1166 D: 3936
sprt @ 15+0.05 th 1 Rewrite the outpost code: chasing the opponent king.
14-05-10 sni outpost5 diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 10607 W: 1928 L: 2005 D: 6674
sprt @ 15+0.05 th 1 Chasing the king : take 2
14-05-10 sni pinned_pawns4 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 27407 W: 4863 L: 4898 D: 17646
sprt @ 15+0.05 th 1 Take 2
14-05-10 sni pinned_pawns4 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 20667 W: 3619 L: 3671 D: 13377
sprt @ 15+0.05 th 1 Take 3
14-05-10 sni outpost5 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 5573 W: 1060 L: 1152 D: 3361
sprt @ 15+0.05 th 1 Chasing the king : take 3 (more bonus for outposts in midgame)
14-05-11 sni combo diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 57134 W: 10131 L: 10091 D: 36912
sprt @ 15+0.05 th 1 Combo patch of two recent patches that failed LTC with positive results : PawnTune_a and pinned_pawns2
14-05-11 sni outpost6 diff
LLR: 2.99 (-2.94,2.94) [-1.50,4.50]
Total: 53666 W: 9759 L: 9517 D: 34390
sprt @ 15+0.05 th 1 My previous attempts tried to chase the king in midgame and endgame. This patch is less radical a change from the official stockfish, and will basically only try to follow the opponent king with pieces in endgame.
14-05-12 sni outpost6 diff
LLR: 2.95 (-2.94,2.94) [0.00,6.00]
Total: 20891 W: 3252 L: 3059 D: 14580
sprt @ 60+0.05 th 1 LTC: my previous attempts tried to chase the king in midgame and endgame. This patch is less radical a change from the official stockfish, and will basically only try to follow the opponent king with pieces in endgame.
14-05-18 sni pinned_pawns5 diff
LLR: -4.91 (-2.94,2.94) [-1.50,4.50]
Total: 26109 W: 4533 L: 4640 D: 16936
sprt @ 15+0.05 th 1 Experimental run: don't get a malus for a piece attacked by a pinned pawn
14-05-18 sni pinned_pawns5 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 8231 W: 1419 L: 1502 D: 5310
sprt @ 15+0.05 th 1 Experimental run: give a bonus for a piece attacked by a pinned pawn
14-05-22 sni rook_passers diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 8048 W: 1375 L: 1459 D: 5214
sprt @ 15+0.05 th 1 Rook passers are strong when the defensive side does not have a double rook.
14-05-23 sni little_combinaisons7 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17813 W: 2917 L: 2977 D: 11919
sprt @ 15+0.05 th 1 Anticipate winning exchange sequences on weak pawns.
14-06-01 sni outpost8 diff
ELO: 1.67 +-1.9 (95%) LOS: 96.1%
Total: 40000 W: 6030 L: 5838 D: 28132
40000 @ 60+0.05 th 1 Experimental run
14-06-17 sni activity diff
ELO: -11.15 +-2.4 (95%) LOS: 0.0%
Total: 24756 W: 3362 L: 4156 D: 17238
40000 @ 60+0.05 th 1 Measure elo at low priority.
14-06-17 sni activity diff
ELO: -4.37 +-1.9 (95%) LOS: 0.0%
Total: 36907 W: 5130 L: 5594 D: 26183
40000 @ 60+0.05 th 1 take 2 : experimental run, low priority
14-07-04 sni tuning_outpost diff
98532/100000 iterations
210000/210000 games played
210000 @ 15+0.05 th 1 Tuning some parameters to try replacement formula instead of the two outpost arrays. Low priority.
14-07-06 sni outpost11 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 2702 W: 434 L: 531 D: 1737
sprt @ 15+0.05 th 1 Try replacement formula instead of the two outpost arrays
14-07-07 sni outpost12 diff
ELO: -0.96 +-2.8 (95%) LOS: 25.3%
Total: 20000 W: 3398 L: 3453 D: 13149
20000 @ 15+0.05 th 1 Measure Elo of some outpost tweaks
14-07-09 sni outpost12 diff
ELO: -0.46 +-1.8 (95%) LOS: 30.9%
Total: 38972 W: 5406 L: 5458 D: 28108
40000 @ 60+0.05 th 1 Experimental run (test promising version after tuning)
14-07-12 sni probcut diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 19250 W: 3168 L: 3225 D: 12857
sprt @ 15+0.05 th 1 Avoid searching the probcut subtrees when the searches are not likely to produce a cut
14-07-12 sni pawnspan diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 47244 W: 7935 L: 7924 D: 31385
sprt @ 15+0.05 th 1 Simplify pawn structure after we have a passed pawn
14-07-12 sni pawnspan diff
LLR: -2.94 (-2.94,2.94) [-1.50,4.50]
Total: 10137 W: 1668 L: 1746 D: 6723
sprt @ 15+0.05 th 1 Take 2
14-07-13 sni pawnspan diff
LLR: 2.96 (-2.94,2.94) [-1.50,4.50]
Total: 18857 W: 3278 L: 3129 D: 12450
sprt @ 15+0.05 th 1 take 3 (PawnsFileSpan = (0,12))
14-07-13 sni pawnspan diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 11912 W: 2015 L: 2089 D: 7808
sprt @ 15+0.05 th 1 take 4 (PawnsFileSpan = (0,18))
14-07-14 sni pawnspan diff
LLR: -2.97 (-2.94,2.94) [0.00,6.00]
Total: 13097 W: 1775 L: 1821 D: 9501
sprt @ 60+0.05 th 1 LTC : take 3 (PawnsFileSpan = (0,12))
14-07-20 sni outpost_table diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 6253 W: 1135 L: 1271 D: 3847
sprt @ 15+0.05 th 1 Use an array as suggested by Lucas to tabulate the formula
14-07-21 sni outpost_table diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 36714 W: 6631 L: 6642 D: 23441
sprt @ 15+0.05 th 1 Take 2
14-07-23 sni isolated diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 16173 W: 2701 L: 2765 D: 10707
sprt @ 15+0.05 th 1 Use !passed instead of opposed condition for isolated pawns
14-07-23 sni pawnspan3 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 15762 W: 2667 L: 2732 D: 10363
sprt @ 15+0.05 th 1 Simplify pawn structure after we get a passed pawn : half bonus for PawnsFileSpan
14-10-02 sni double_isolated diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 30966 W: 6258 L: 6279 D: 18429
sprt @ 15+0.05 th 1 Full penalties for isolated doubled pawns and backward doubled pawns (even if the pawns are opposed)
14-10-02 sni more_connections diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 2834 W: 578 L: 680 D: 1576
sprt @ 15+0.05 th 1 Count more connections between pawns : the aim is to get more continuity in the eval when a pawn duet (say, b4-c4) is pushed.
14-10-03 sni double_isolated diff
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 16423 W: 3424 L: 3270 D: 9729
sprt @ 15+0.05 th 1 Take 2 : full penalties for isolated pawns and backward pawns (if the pawns have any other weakness).
14-10-03 sni double_isolated diff
LLR: -2.95 (-2.94,2.94) [0.00,6.00]
Total: 15161 W: 2562 L: 2591 D: 10008
sprt @ 60+0.05 th 1 LTC for take 2 : full penalties for isolated pawns and backward pawns (if the pawns have any other weakness).
14-10-04 sni doubled_isolated diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 5583 W: 1100 L: 1192 D: 3291
sprt @ 15+0.05 th 1 Take 3 : only the case of doubled isolated pawns
14-10-05 sni keep_material4 diff
LLR: -2.96 (-2.94,2.94) [-1.50,4.50]
Total: 12668 W: 2546 L: 2618 D: 7504
sprt @ 15+0.05 th 1 Keep more pawns on the board when we are ahead
14-10-07 sni keep_material5 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 17880 W: 3615 L: 3672 D: 10593
sprt @ 15+0.05 th 1 Keep more pawns on the board when we are ahead: take 2.
14-10-07 sni passed_duo5 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 4586 W: 918 L: 1013 D: 2655
sprt @ 15+0.05 th 1 Add some bonus for each duo of passed pawns
14-10-07 sni iterative_deepening diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 3110 W: 598 L: 697 D: 1815
sprt @ 15+0.05 th 1 Search even depths in iterative deepening
14-10-07 sni doubled_isolated2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 28982 W: 5890 L: 5916 D: 17176
sprt @ 15+0.05 th 1 Increase penalty for doubled pawn if pawn is also isolated
14-10-08 sni doubled_isolated2 diff
4957/5000 iterations
10000/10000 games played
10000 @ 15+0.05 th 1 SPSA tuning for isolated doubled penalty
14-10-08 sni doubled_isolated2 diff
LLR: -2.95 (-2.94,2.94) [-1.50,4.50]
Total: 21464 W: 4355 L: 4402 D: 12707
sprt @ 15+0.05 th 1 sprt test with penalty for doubled isolated pawns tuned by SPSA
14-10-08 sni keep_material6 diff
LLR: -2.97 (-2.94,2.94) [-1.50,4.50]
Total: 14959 W: 3001 L: 3067 D: 8891
sprt @ 15+0.05 th 1 Take 3 : when ahead, keep pawns and exchange pieces