| 16-10-11 |
sni |
draw_value |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 21037 W: 3853 L: 3888 D: 13296
|
sprt
@ 10+0.1 th 1 |
Try decreasing contempt with decreasing piece count |
| 16-10-11 |
luc |
7th_rank_outposts |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 18242 W: 3231 L: 3325 D: 11686
|
sprt
@ 10+0.1 th 1 |
Include 7th rank in outpost squares |
| 16-10-12 |
Voy |
qst |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 11973 W: 2145 L: 2219 D: 7609
|
sprt
@ 10+0.1 th 1 |
Don't evasion prune at root of QS. |
| 16-10-12 |
atu |
is_endgame |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 6788 W: 1192 L: 1288 D: 4308
|
sprt
@ 10+0.1 th 1 |
KingSafetyEvalThreshold = 11 |
| 16-10-11 |
atu |
tune_king_safety |
diff
|
9772/10000 iterations
20000/20000 games played
|
20000
@ 20+0.2 th 1 |
Tune the new KingSafetyEvalThreshold parameter (sorry, fixed wrong version) |
| 16-10-10 |
Voy |
red2d |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 65085 W: 11499 L: 11437 D: 42149
|
sprt
@ 10+0.1 th 1 |
Last shot before I wave the white flag... |
| 16-10-10 |
Roc |
ContactChecks |
diff
|
LLR: 2.97 (-2.94,2.94) [0.00,5.00]
Total: 61534 W: 11028 L: 10646 D: 39860
|
sprt
@ 10+0.1 th 1 |
Compute a few missing contact check bonus
|
| 16-10-10 |
syz |
futility_revert |
diff
|
LLR: 2.95 (-2.94,2.94) [-3.00,1.00]
Total: 48728 W: 8654 L: 8580 D: 31494
|
sprt
@ 10+0.1 th 1 |
Find out whether correcting the futility pruning logic loses Elo (official = SF master). See discussion https://github.com/official-stockfish/Stockfish/commit/f799610d4bb48bc280ea7f58cd5f78ab21028bf5 |
| 16-10-11 |
Voy |
nullCM |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 27283 W: 4904 L: 4913 D: 17466
|
sprt
@ 10+0.1 th 1 |
My take on Null - CM |
| 16-10-11 |
Elb |
rm_tweak |
diff
|
LLR: -3.29 (-2.94,2.94) [0.00,4.00]
Total: 22505 W: 3991 L: 4089 D: 14425
|
sprt
@ 10+0.1 th 1 |
razor_margin tweak |
| 16-10-10 |
atu |
is_endgame |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13656 W: 2400 L: 2467 D: 8789
|
sprt
@ 10+0.1 th 1 |
Skip evaluation of king safety (and other middlegame-specific features) closer to the endgame |
| 16-10-10 |
sni |
snipers |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12495 W: 2215 L: 2287 D: 7993
|
sprt
@ 10+0.1 th 1 |
Take 4, bonus = S(15, 0) |
| 16-10-10 |
luc |
simple_psqt |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 7571 W: 1295 L: 1426 D: 4850
|
sprt
@ 10+0.1 th 1 |
Still no endgame bonus, but bump up midgame values for outposts... |
| 16-10-10 |
sni |
snipers |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7439 W: 1296 L: 1390 D: 4753
|
sprt
@ 10+0.1 th 1 |
Take 3, bonus = S(25, 0) |
| 16-10-10 |
Voy |
PGO |
diff
|
ELO: 0.96 +-2.9 (95%) LOS: 74.4%
Total: 20000 W: 3541 L: 3486 D: 12973
|
20000
@ 10+0.1 th 1 |
Compare PGO with default bench depth 13 to depth 14. My concern is that pruning patches may have a disadvantage...as it will send less time being optimized. |
| 16-10-10 |
Roc |
HangingAndQueen |
diff
|
LLR: 2.95 (-2.94,2.94) [0.00,5.00]
Total: 12840 W: 2354 L: 2176 D: 8310
|
sprt
@ 10+0.1 th 1 |
Consider hanging if only defended by a queen and attacked by 2 |
| 16-10-10 |
sni |
snipers |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 12332 W: 2127 L: 2200 D: 8005
|
sprt
@ 10+0.1 th 1 |
Bonus for counter attacks on snipers. Take 2, bonus = S(30,30) |
| 16-10-10 |
Roc |
QueenChecks |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 13506 W: 2341 L: 2409 D: 8756
|
sprt
@ 10+0.1 th 1 |
Adjust the QueenCheck bonus according to distance to King. |
| 16-10-10 |
sni |
snipers |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 34218 W: 6109 L: 6090 D: 22019
|
sprt
@ 10+0.1 th 1 |
Bonus for counter attacks on snipers. Take 1, bonus = S(35,0) |
| 16-10-10 |
fau |
Newpatch |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 7819 W: 1334 L: 1426 D: 5059
|
sprt
@ 10+0.1 th 1 |
2 different bonuses for rookonpawn |
| 16-10-10 |
syz |
cm_nullmove |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 4284 W: 672 L: 778 D: 2834
|
sprt
@ 10+0.1 th 1 |
Set ss->counterMoves also for null moves and avoid nullptr checks. Corrected version. (official = SF master) |
| 16-10-09 |
Voy |
red2c |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 64157 W: 8237 L: 8149 D: 47771
|
sprt
@ 60+0.6 th 1 |
With a smaller denominator...I believe this may fare well at LTC. |
| 16-10-09 |
fau |
tune |
diff
|
29189/30000 iterations
59262/60000 games played
|
60000
@ 20+0.2 th 1 |
Tuning an idea which I'm testing locally with good potential |
| 16-10-10 |
sg |
pruning |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 6983 W: 1183 L: 1279 D: 4521
|
sprt
@ 10+0.1 th 1 |
Don't do move count pruning if in check |
| 16-10-09 |
Elb |
block_them |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 18078 W: 3177 L: 3226 D: 11675
|
sprt
@ 10+0.1 th 1 |
Take 3 |
| 16-10-09 |
sg |
pruning |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 33863 W: 5970 L: 5953 D: 21940
|
sprt
@ 10+0.1 th 1 |
Don't prune captures if in check |
| 16-10-09 |
jos |
cm_tweak |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 23956 W: 4207 L: 4282 D: 15467
|
sprt
@ 10+0.1 th 1 |
Change penalty only. |
| 16-10-09 |
SC |
stonewallInit |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 21274 W: 3754 L: 3790 D: 13730
|
sprt
@ 10+0.1 th 1 |
I learned that there is some information hidden in this stonewall stuff. Is classical stonewall good for blocking positions? |
| 16-10-09 |
SC |
stonewallInit |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 16405 W: 2935 L: 2990 D: 10480
|
sprt
@ 10+0.1 th 1 |
Or do they favour strong side? |
| 16-10-09 |
jos |
cm_tweak |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,4.00]
Total: 8783 W: 1501 L: 1627 D: 5655
|
sprt
@ 10+0.1 th 1 |
Stats tweak. |
| 16-10-09 |
mco |
pieces_no_type |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 5329 W: 893 L: 995 D: 3441
|
sprt
@ 10+0.1 th 1 |
Remove byTypeBB and byColorBB[]: speed optimization test |
| 16-10-09 |
Elb |
block_them |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 37648 W: 6668 L: 6635 D: 24345
|
sprt
@ 10+0.1 th 1 |
Take 2: increase bonus for blocking unsupported opposite pawns |
| 16-10-08 |
Voy |
red2b |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 21421 W: 2747 L: 2796 D: 15878
|
sprt
@ 60+0.6 th 1 |
See if this will scale well at LTC... (lower throughput) |
| 16-10-08 |
Voy |
red2c |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 35746 W: 6363 L: 6338 D: 23045
|
sprt
@ 10+0.1 th 1 |
take 3... keep the denominator as 2. |
| 16-10-08 |
aji |
fpeval_simplify |
diff
|
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 26905 W: 3503 L: 3390 D: 20012
|
sprt
@ 60+0.6 th 1 |
Do we need to tweak the eval during futlity pruning? Test as simplification: LTC |
| 16-10-08 |
fau |
rookonpawn |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,4.00]
Total: 14590 W: 2592 L: 2699 D: 9299
|
sprt
@ 10+0.1 th 1 |
Rook On Pawn Tweak |
| 16-10-08 |
fau |
test1 |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,4.00]
Total: 38568 W: 6848 L: 6874 D: 24846
|
sprt
@ 10+0.1 th 1 |
Phalanxes 5/9 |
| 16-10-08 |
Elb |
block_them |
diff
|
LLR: -2.97 (-2.94,2.94) [0.00,5.00]
Total: 17440 W: 3070 L: 3122 D: 11248
|
sprt
@ 10+0.1 th 1 |
MinorBehindPawn tweak: bonus for blocking opposing pawns on all ranks |
| 16-10-08 |
sni |
adjacent_bishops_attack |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 20240 W: 3536 L: 3576 D: 13128
|
sprt
@ 10+0.1 th 1 |
Take 4: Use all their camp |
| 16-10-08 |
fau |
test2 |
diff
|
ELO: -2.37 +-3.7 (95%) LOS: 10.5%
Total: 12000 W: 2107 L: 2189 D: 7704
|
12000
@ 10+0.1 th 1 |
Phalanxes test |
| 16-10-08 |
lan |
advanced_pawn_push |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 6232 W: 1117 L: 1216 D: 3899
|
sprt
@ 10+0.1 th 1 |
Tweak advanced_pawn_push |
| 16-10-08 |
aji |
nmp_tweak |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 1339 W: 189 L: 309 D: 841
|
sprt
@ 10+0.1 th 1 |
NMP tweak: STC
trying to make sense out of the staticEval check |
| 16-10-08 |
sni |
cooperative_knights |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8673 W: 1500 L: 1588 D: 5585
|
sprt
@ 10+0.1 th 1 |
Take 2: use their whole camp |
| 16-10-08 |
aji |
fpeval_simplify |
diff
|
LLR: 2.96 (-2.94,2.94) [-3.00,1.00]
Total: 167687 W: 29778 L: 29904 D: 108005
|
sprt
@ 10+0.1 th 1 |
Do we need to tweak the eval during futlity pruning? Test as simplification: STC |
| 16-10-08 |
sni |
adjacent_bishops_attack |
diff
|
LLR: -2.96 (-2.94,2.94) [0.00,5.00]
Total: 25427 W: 4517 L: 4535 D: 16375
|
sprt
@ 10+0.1 th 1 |
Take 3, bonus = S(14,0) |
| 16-10-08 |
sni |
cooperative_knights |
diff
|
LLR: -2.94 (-2.94,2.94) [0.00,5.00]
Total: 12963 W: 2256 L: 2326 D: 8381
|
sprt
@ 10+0.1 th 1 |
Bonus for good knight coordination |
| 16-10-08 |
sni |
adjacent_bishops_attack |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 7710 W: 1298 L: 1390 D: 5022
|
sprt
@ 10+0.1 th 1 |
Take 2, bonus = S(50, 0) |
| 16-10-08 |
Elb |
block_weak2 |
diff
|
Pending...
|
sprt
@ 10+0.1 th 1 |
MinorBehindPawn tweak: extend ranks for opposing weak pawns (fixed bench) |
| 16-10-08 |
SC |
stonewallSN |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 8582 W: 1032 L: 1122 D: 6428
|
sprt
@ 60+0.6 th 1 |
Stonewall only if opponents ha's not stopped the rear pawn. After that I will again wait sometime before going on... LTC |
| 16-10-08 |
Voy |
red2b |
diff
|
LLR: -2.95 (-2.94,2.94) [0.00,5.00]
Total: 53911 W: 9622 L: 9520 D: 34769
|
sprt
@ 10+0.1 th 1 |
Take 2 |