Stockfish Testing Queue

Pending - 1 tests 0.0 hrs

18-11-21 Kur cttime6 diff
Pending... (0.7 hrs)
[0.00,5.00]
sprt @ 60+0.6 th 1 Satisfied with STC results which show no change in elo - this experiment is aimed at longer TC and should hardly change STC. Priority -1.

Active - 0 tests

Finished - 206 tests

18-11-21 Kur cttime6 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19583 W: 4323 L: 4358 D: 10902
sprt @ 10+0.1 th 1 Don't increase contempt much even on low time.
18-11-20 Kur cttime4 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12630 W: 2713 L: 2782 D: 7135
sprt @ 10+0.1 th 1 Relate contempt solely to time rather than time advantage.
18-11-20 Kur restricted diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11659 W: 2508 L: 2582 D: 6569
sprt @ 10+0.1 th 1 Don't count squares with pieces as restricted.
18-11-20 Kur trappedrook3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7691 W: 1708 L: 1802 D: 4181
sprt @ 10+0.1 th 1 Give penalty regardless of king position if rook is in the corner.
18-11-20 Kur trappedrook3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 3927 W: 826 L: 939 D: 2162
sprt @ 10+0.1 th 1 Give trappedrook penalty regardless of king position when on rank 1.
18-11-19 Kur cttime3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 26895 W: 5878 L: 5877 D: 15140
sprt @ 10+0.1 th 1 Take 2 (fixed).
18-11-19 Kur cttime diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 22382 W: 4703 L: 4726 D: 12953
sprt @ 10+0.1 th 1 Relate contempt to time.
18-11-18 Kur dblext diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6283 W: 1297 L: 1397 D: 3589
sprt @ 10+0.1 th 1 Extend 2 plies for castling.
18-11-17 Kur hemmedbishop diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9332 W: 1931 L: 2016 D: 5385
sprt @ 10+0.1 th 1 Bishop penalty, fixed again. (Thanks Viz and Proton)
18-11-17 Kur trappedrook diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 25353 W: 5401 L: 5409 D: 14543
sprt @ 10+0.1 th 1 Penalty for a trapped rook stuck behind a trapped piece.
18-11-17 Kur bishoppawns diff
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 10150 W: 2162 L: 2244 D: 5744
sprt @ 10+0.1 th 1 (25, 25).
18-11-17 Kur bishoppawns diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 33370 W: 7178 L: 7147 D: 19045
sprt @ 10+0.1 th 1 Penalty when our bishop can't move due to opposing pawn control.
18-11-16 Kur a6h6tweak diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 38960 W: 8371 L: 8313 D: 22276
sprt @ 10+0.1 th 1 A few asymmetric parameter tweaks.
18-11-16 Kur kingpiece6 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 8429 W: 1778 L: 1868 D: 4783
sprt @ 10+0.1 th 1 Exclude knights.
18-11-16 Kur blockedrook diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 13649 W: 2873 L: 2938 D: 7838
sprt @ 10+0.1 th 1 Require rank 1.
18-11-16 Kur simptr diff
LLR: -2.95 (-2.94,2.94) [-3.00,1.00]
Total: 3709 W: 755 L: 928 D: 2026
sprt @ 10+0.1 th 1 Does the king position requirement here actually help?
18-11-16 Kur kingpiece diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 17466 W: 3712 L: 3758 D: 9996
sprt @ 10+0.1 th 1 (10, 0).
18-11-16 Kur blockedrook diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 24950 W: 5245 L: 5256 D: 14449
sprt @ 10+0.1 th 1 Rook behind pawn penalty.
18-11-16 Kur kingpiece3 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 30738 W: 6584 L: 6566 D: 17588
sprt @ 10+0.1 th 1 Apply bonus for queenside as well.
18-11-16 Kur kingpiece5 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 16310 W: 3467 L: 3519 D: 9324
sprt @ 10+0.1 th 1 Apply bonus for advanced pieces as well.
18-11-16 Kur kingpiece diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25728 W: 5451 L: 5458 D: 14819
sprt @ 10+0.1 th 1 Bigger bonus (6, 0).
18-11-16 Kur kingpiece diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 35117 W: 7419 L: 7381 D: 20317
sprt @ 10+0.1 th 1 Kingside piece bonus.
18-11-15 Kur conconsistancy diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 11692 W: 2480 L: 2554 D: 6658
sprt @ 10+0.1 th 1 Last try unless promising - c30.
18-11-15 Kur conconsistancy diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 11821 W: 2469 L: 2542 D: 6810
sprt @ 10+0.1 th 1 Interpolate contempt - c24.
18-11-15 Kur conconsistancy diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18772 W: 3844 L: 3885 D: 11043
sprt @ 10+0.1 th 1 Interpolate contempt.
18-11-11 Kur tunedeval diff
LLR: 2.95 (-2.94,2.94) [0.00,4.00]
Total: 157625 W: 25533 L: 24893 D: 107199
sprt @ 60+0.6 th 1 Tuned eval scores.
18-11-14 Kur lazythres diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 27697 W: 5936 L: 5992 D: 15769
sprt @ 10+0.1 th 1 Increase lazy threshold.
18-11-14 Kur trappedking diff
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 18186 W: 3887 L: 3929 D: 10370
sprt @ 10+0.1 th 1 Penalty for having king trapped in the corner.
18-11-13 Kur tunedeval diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 84697 W: 18173 L: 18009 D: 48515
sprt @ 10+0.1 th 1 STC: tuned eval scores.
18-11-10 Kur tuneeval diff
48133/50000 iterations
100000/100000 games played
100000 @ 60+0.6 th 1 Tune the list of evaluation scores.
18-11-11 Kur pawncheck diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 24918 W: 5369 L: 5379 D: 14170
sprt @ 10+0.1 th 1 Bigger bonus (700).
18-11-11 Kur pawncheck diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 31688 W: 6797 L: 6774 D: 18117
sprt @ 10+0.1 th 1 Bigger penalty (600).
18-11-10 Kur pawncheck2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25093 W: 5292 L: 5302 D: 14499
sprt @ 10+0.1 th 1 Slightly bigger penalty.
18-11-10 Kur pawncheck diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 10435 W: 2163 L: 2243 D: 6029
sprt @ 10+0.1 th 1 Try smaller penalty.
18-11-10 Kur pawncheck diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 21551 W: 4606 L: 4632 D: 12313
sprt @ 10+0.1 th 1 SafeCheck bonus for pawns (fixed).
18-11-09 Kur pawncheck diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 776 W: 131 L: 263 D: 382
sprt @ 10+0.1 th 1 SafeCheck bonus for pawns.
18-11-10 Kur spaceweight diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7411 W: 1567 L: 1662 D: 4182
sprt @ 10+0.1 th 1 Adjust weighting for space eval.
18-11-09 Kur tuned8x8kpsqt100k diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 30688 W: 4899 L: 4903 D: 20886
sprt @ 60+0.6 th 1 Last LTC attempt for this version. Attempt to further narrow ELO range.
18-11-08 Kur tuned8x8kpsqt diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 23688 W: 3821 L: 3851 D: 16016
sprt @ 60+0.6 th 1 See if a more convincing result can be achieved with the fully tuned values.
18-11-08 Kur tunedknights2 diff
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 20797 W: 3262 L: 3351 D: 14184
sprt @ 60+0.6 th 1 1/2 changes needed, as expected.
18-11-08 Kur tunedknights diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 11791 W: 1831 L: 1947 D: 8013
sprt @ 60+0.6 th 1 Knight PSQT and mobility tuned. Values changed a *lot* so will likely need to try 1/2 or 1/3 at some point. Consensus on github seems to be to test LTC tunes directly at LTC (makes 100% sense to me). Using tp 100 since details have not been worked out (personally I think tp100/166 is best for these LTC-tuned-value tests).
18-11-07 Kur tuned8x8kpsqt100k diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 61662 W: 10011 L: 9893 D: 41758
sprt @ 60+0.6 th 1 Run 2nd LTC to verify ELO gain.
18-11-08 Kur rllll1 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 6848 W: 1430 L: 1528 D: 3890
sprt @ 10+0.1 th 1 Don't give the bonus when on a semi-open file.
18-10-31 Kur ltcknighttune2 diff
76995/80000 iterations
160001/160000 games played
160000 @ 60+0.6 th 1 LTC knight tuning: resubmitted.
18-10-29 Kur 8x8kingPSQTtune diff
72178/75000 iterations
150001/150000 games played
150000 @ 60+0.6 th 1 Tune experimental asymmetric king PQST. tp=250, priority -1. (Please check if I did everything right/values are set optimally).
18-11-06 Kur tuned8x8kpsqt100k diff
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 28178 W: 4625 L: 4394 D: 19159
sprt @ 60+0.6 th 1 Fishtest is relatively empty so try this at LTC since that's where it was tuned. tp 100.
18-11-06 Kur trrrr2 diff
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 18061 W: 3895 L: 3988 D: 10178
sprt @ 10+0.1 th 1 Give trappedrook penalty more sparingly (mob<=2).
18-11-05 Kur trrrr1 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6460 W: 1329 L: 1428 D: 3703
sprt @ 10+0.1 th 1 Testing at [0,5] because this shouldn't be removed unless it gains elo. (I'd say [0,4] is optimal but it's not a p-tweak).
18-11-05 Kur trappedrook2 diff
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 19148 W: 4090 L: 4128 D: 10930
sprt @ 10+0.1 th 1 Simple small penalty for bad h-file rooks.
18-11-05 Kur rookmob3 diff
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 9344 W: 1975 L: 2060 D: 5309
sprt @ 10+0.1 th 1 Further penalty for low mobility rooks on the edge.