WDL

[zullil]

As many of know, NN engines have been reporting WDL data with Scoring Percentage for some time. A while back, I took their formula to convert to a centipawn and applied it the centipawn evaluation to come u

It is worth noting that Stockfish's WDL is a function of both the evaluation and the current game ply.

https://github.com/official-stockfish/S ... dfdf670392

That makes it useless for analysis of positions or studies, for which the ply is immaterial or unknown. It would be extremely useful to have a wdl feature that accounted not for "ply" but for the material on the board (for which "ply" is likely a proxy) and whether the value derives from a TB (or some other known certain draw such as perpetual).

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

Though it drops to about 4% if we're at move 120 of the game.

Code: Select all

$ ./stockfish 
Stockfish 010720 64 BMI2 by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
uci
id name Stockfish 010720 64 BMI2
id author T. Romstad, M. Costalba, J. Kiiski, G. Linscott

option name Debug Log File type string default 
option name Contempt type spin default 24 min -100 max 100
option name Analysis Contempt type combo default Both var Off var White var Black var Both
option name Threads type spin default 1 min 1 max 512
option name Hash type spin default 16 min 1 max 33554432
option name Clear Hash type button
option name Ponder type check default false
option name MultiPV type spin default 1 min 1 max 500
option name Skill Level type spin default 20 min 0 max 20
option name Move Overhead type spin default 10 min 0 max 5000
option name Slow Mover type spin default 100 min 10 max 1000
option name nodestime type spin default 0 min 0 max 10000
option name UCI_Chess960 type check default false
option name UCI_AnalyseMode type check default false
option name UCI_LimitStrength type check default false
option name UCI_Elo type spin default 1350 min 1350 max 2850
option name UCI_ShowWDL type check default true
option name SyzygyPath type string default <empty>
option name SyzygyProbeDepth type spin default 1 min 1 max 100
option name Syzygy50MoveRule type check default true
option name SyzygyProbeLimit type spin default 7 min 0 max 7
uciok
setoption name SyzygyPath value /home/louis/Documents/Chess/tb/src
info string Found 510 tablebases
position fen 4k3/8/8/8/8/8/8/4K3 w - - 0 1
d

 +---+---+---+---+---+---+---+---+
 |   |   |   |   | k |   |   |   | 8
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 7
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 6
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 5
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 4
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 3
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 2
 +---+---+---+---+---+---+---+---+
 |   |   |   |   | K |   |   |   | 1
 +---+---+---+---+---+---+---+---+
   a   b   c   d   e   f   g   h

Fen: 4k3/8/8/8/8/8/8/4K3 w - - 0 1
Key: 279216B9424835AA
Checkers: 
Tablebases WDL: Draw (Success)
Tablebases DTZ:    0 (Success)
go depth 1
info depth 1 seldepth 1 multipv 1 score cp 0 wdl 106 788 106 nodes 5 nps 5000 tbhits 5 time 1 pv e1d1
bestmove e1d1
position fen 4k3/8/8/8/8/8/8/4K3 w - - 0 240
d

 +---+---+---+---+---+---+---+---+
 |   |   |   |   | k |   |   |   | 8
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 7
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 6
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 5
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 4
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 3
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 2
 +---+---+---+---+---+---+---+---+
 |   |   |   |   | K |   |   |   | 1
 +---+---+---+---+---+---+---+---+
   a   b   c   d   e   f   g   h

Fen: 4k3/8/8/8/8/8/8/4K3 w - - 0 240
Key: 279216B9424835AA
Checkers: 
Tablebases WDL: Draw (Success)
Tablebases DTZ:    0 (Success)
go depth 1
info depth 1 seldepth 1 multipv 1 score cp 0 wdl 38 924 38 nodes 5 nps 2500 tbhits 5 time 2 pv e1d1
bestmove e1d1

[Vinvin]

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

The formula seems not very accurate yet

[zullil]

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

The formula seems not very accurate yet

Lc0 knows better:

info depth 1 seldepth 2 time 2784 nodes 261 score cp 0 wdl 0 1000 0 nps 52200 tbhits 1 pv e1d1

[Milos]

As many of know, NN engines have been reporting WDL data with Scoring Percentage for some time. A while back, I took their formula to convert to a centipawn and applied it the centipawn evaluation to come u

It is worth noting that Stockfish's WDL is a function of both the evaluation and the current game ply.

https://github.com/official-stockfish/S ... dfdf670392

That makes it useless for analysis of positions or studies, for which the ply is immaterial or unknown. It would be extremely useful to have a wdl feature that accounted not for "ply" but for the material on the board (for which "ply" is likely a proxy) and whether the value derives from a TB (or some other known certain draw such as perpetual).

Yeah, ply is never a good indicator of anything, like game phase. Usually in closed positions with a ton of shuffling engines that use ply instead of material to calculate game phase fail miserably.

[Ovyron]

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

Though it drops to about 4% if we're at move 120 of the game.

I've always been against WDL, saying why it's either inaccurate or misleading for most scenarios, I like how this example just showcases what you're expected to see in reality.

I still can't wrap my head around something like "White has a 50% chance to win this", because for that to be useful I also need to know black's winning chances. If black has 0% chance to win this then this position is a great one to aim for. If black also has 50% chance to win this then the expected performance is 50%, and I'd rather play into one where White's chances are only 30% but black's are only 10%.

But WDL has no way to show that difference, so it falls flat on its face.

[zullil]

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

Though it drops to about 4% if we're at move 120 of the game.

I've always been against WDL, saying why it's either inaccurate or misleading for most scenarios, I like how this example just showcases what you're expected to see in reality.

I still can't wrap my head around something like "White has a 50% chance to win this", because for that to be useful I also need to know black's winning chances. If black has 0% chance to win this then this position is a great one to aim for. If black also has 50% chance to win this then the expected performance is 50%, and I'd rather play into one where White's chances are only 30% but black's are only 10%.

But WDL has no way to show that difference, so it falls flat on its face.

If WDL shows 200 500 300, for example, then the interpretation is that the side-to move has a 20% chance of winning and a 30% chance of losing. So the opponent has a 30% chance of winning and a 20% chance of losing. There are only three outcomes:

1. The side-to-move wins (so the opponent loses).
2. The game is drawn.
3. The side-to-move loses (so the opponent wins).

[chrisw]

As many of know, NN engines have been reporting WDL data with Scoring Percentage for some time. A while back, I took their formula to convert to a centipawn and applied it the centipawn evaluation to come u

It is worth noting that Stockfish's WDL is a function of both the evaluation and the current game ply.

https://github.com/official-stockfish/S ... dfdf670392

That makes it useless for analysis of positions or studies, for which the ply is immaterial or unknown. It would be extremely useful to have a wdl feature that accounted not for "ply" but for the material on the board (for which "ply" is likely a proxy) and whether the value derives from a TB (or some other known certain draw such as perpetual).

Here's a fun edge case:

[d]4k3/8/8/8/8/8/8/4K3 w - - 0 1

About an 11% chance that White wins this, according to Stockfish.

Though it drops to about 4% if we're at move 120 of the game.

Code: Select all

$ ./stockfish 
Stockfish 010720 64 BMI2 by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
uci
id name Stockfish 010720 64 BMI2
id author T. Romstad, M. Costalba, J. Kiiski, G. Linscott

option name Debug Log File type string default 
option name Contempt type spin default 24 min -100 max 100
option name Analysis Contempt type combo default Both var Off var White var Black var Both
option name Threads type spin default 1 min 1 max 512
option name Hash type spin default 16 min 1 max 33554432
option name Clear Hash type button
option name Ponder type check default false
option name MultiPV type spin default 1 min 1 max 500
option name Skill Level type spin default 20 min 0 max 20
option name Move Overhead type spin default 10 min 0 max 5000
option name Slow Mover type spin default 100 min 10 max 1000
option name nodestime type spin default 0 min 0 max 10000
option name UCI_Chess960 type check default false
option name UCI_AnalyseMode type check default false
option name UCI_LimitStrength type check default false
option name UCI_Elo type spin default 1350 min 1350 max 2850
option name UCI_ShowWDL type check default true
option name SyzygyPath type string default <empty>
option name SyzygyProbeDepth type spin default 1 min 1 max 100
option name Syzygy50MoveRule type check default true
option name SyzygyProbeLimit type spin default 7 min 0 max 7
uciok
setoption name SyzygyPath value /home/louis/Documents/Chess/tb/src
info string Found 510 tablebases
position fen 4k3/8/8/8/8/8/8/4K3 w - - 0 1
d

 +---+---+---+---+---+---+---+---+
 |   |   |   |   | k |   |   |   | 8
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 7
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 6
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 5
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 4
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 3
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 2
 +---+---+---+---+---+---+---+---+
 |   |   |   |   | K |   |   |   | 1
 +---+---+---+---+---+---+---+---+
   a   b   c   d   e   f   g   h

Fen: 4k3/8/8/8/8/8/8/4K3 w - - 0 1
Key: 279216B9424835AA
Checkers: 
Tablebases WDL: Draw (Success)
Tablebases DTZ:    0 (Success)
go depth 1
info depth 1 seldepth 1 multipv 1 score cp 0 wdl 106 788 106 nodes 5 nps 5000 tbhits 5 time 1 pv e1d1
bestmove e1d1
position fen 4k3/8/8/8/8/8/8/4K3 w - - 0 240
d

 +---+---+---+---+---+---+---+---+
 |   |   |   |   | k |   |   |   | 8
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 7
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 6
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 5
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 4
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 3
 +---+---+---+---+---+---+---+---+
 |   |   |   |   |   |   |   |   | 2
 +---+---+---+---+---+---+---+---+
 |   |   |   |   | K |   |   |   | 1
 +---+---+---+---+---+---+---+---+
   a   b   c   d   e   f   g   h

Fen: 4k3/8/8/8/8/8/8/4K3 w - - 0 240
Key: 279216B9424835AA
Checkers: 
Tablebases WDL: Draw (Success)
Tablebases DTZ:    0 (Success)
go depth 1
info depth 1 seldepth 1 multipv 1 score cp 0 wdl 38 924 38 nodes 5 nps 2500 tbhits 5 time 2 pv e1d1
bestmove e1d1

Plus LC0 knows immediately.
What does this prove? That with one PV that returns 0 from leaf to root, one can prove nothing.
With thousands of PVs all returning zero, one can begin to draw (sic) conclusions.
Which suggests, with multi-PV all returning 0, an AB program could begin to draw concludions.
Ah, got it! If AB returns draw=0, set the window on all other moves to draw-1, then repeated “mind changing” at the root would indicate an actual draw.

[MikeB]

All good comments - as an FYI, this is the current implementation by dev-Stockfish on this position.

This is a known draw position:

[d]8/1p2KP2/1p4q1/1Pp5/2P5/N1Pp1k2/3P4/1N6 b - - 76 40

# stockfish
Stockfish 010720 64 POPCNT by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
position fen 8/1p2KP2/1p4q1/1Pp5/2P5/N1Pp1k2/3P4/1N6 b - - 76 40
setoption name threads value 60
info string Hash table allocation: Windows large pages used.
setoption name hash value 86688
info string Hash table allocation: Windows large pages used.
go depth 50
info depth 1 seldepth 2 multipv 1 score cp 41 wdl 123 860 17 nodes 7985 nps 1140714 tbhits 0 time 7 pv g6g7
info depth 2 seldepth 2 multipv 1 score cp 49 wdl 146 840 14 nodes 22932 nps 2866500 tbhits 0 time 8 pv g6g7 e7e8
info depth 3 seldepth 6 multipv 1 score cp 134 wdl 606 392 2 nodes 52666 nps 5851777 tbhits 0 time 9 pv g6e4 e7d6 e4f4 d6e7 f4e5 e7f8
info depth 4 seldepth 8 multipv 1 score cp 44 wdl 131 853 16 nodes 80073 nps 8897000 tbhits 0 time 9 pv g6g7 e7e8 g7e5 e8d7 e5f6
info depth 5 seldepth 7 multipv 1 score cp 32 wdl 101 878 21 nodes 94836 nps 9483600 tbhits 0 time 10 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 6 seldepth 10 multipv 1 score cp 31 wdl 98 880 22 nodes 103241 nps 10324100 tbhits 0 time 10 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8 f6e6 e8f8 e6e5 f8g8 e5g3 g8h7
info depth 7 seldepth 12 multipv 1 score cp 27 wdl 89 887 24 nodes 129708 nps 12970800 tbhits 0 time 10 pv g6h7 e7e8 h7e4 e8f8 e4g6 f8e8 f3f4
info depth 8 seldepth 10 multipv 1 score cp 269 wdl 980 20 0 nodes 142867 nps 14286700 tbhits 0 time 10 pv g6h7 e7e8 h7h5 e8e7 h5g5 e7f8 f3f4
info depth 9 seldepth 14 multipv 1 score cp 370 wdl 998 2 0 nodes 168957 nps 15359727 tbhits 0 time 11 pv g6h7 e7e8 h7h5 e8e7 h5g5 e7e8 g5e5 e8d7 e5f6
info depth 10 seldepth 22 multipv 1 score cp 370 wdl 998 2 0 nodes 264608 nps 20354461 tbhits 0 time 13 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3e2 a7b6 e2e1 b6a5 e7c7 b5b6 c7d8
info depth 11 seldepth 18 multipv 1 score cp 408 wdl 999 1 0 nodes 299464 nps 21390285 tbhits 0 time 14 pv g6g5 e7f8 g5f6 f8g8 f6g6 g8f8 g6g5 f8e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7
info depth 12 seldepth 27 multipv 1 score cp 393 wdl 999 1 0 nodes 638940 nps 37584705 tbhits 0 time 17 pv g6g7 e7e8 g7e5 e8d8 e5f6 d8e8 f6e6 e8f8 f3f4 f8g7 e6g4 g7h8 g4f5 h8g7 f5g5 g7f8 g5f6
info depth 13 seldepth 25 multipv 1 score cp 428 wdl 1000 0 0 nodes 795859 nps 44214388 tbhits 0 time 18 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3f2 a7b6 f2e1 b6a5 e7f6 a5a4 e1e2 a4b3 b7b6 b3b2 f6h6 b2b3 h6f4
info depth 14 seldepth 29 multipv 1 score cp 397 wdl 999 1 0 nodes 1237370 nps 56244090 tbhits 0 time 22 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 e7c7 a5a4 f3e2 a4b3 c7d7 b3a2 b7b6 a2b3 d7e8 b3a2 e8e4 a2b2 e4e7 b2b3
info depth 15 seldepth 31 multipv 1 score cp 397 wdl 999 1 0 nodes 1398900 nps 60821739 tbhits 0 time 23 pv g6h7 e7e8 h7h5 e8f8 f3f4 f8e7 h5e5 e7d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 f4f3 a5a4 f3e2 a4b3 e7d7 b3b2 d7e8 b2a2 b7b6 a2b3
info depth 16 seldepth 32 multipv 1 score cp 397 wdl 999 1 0 nodes 1429623 nps 59567625 tbhits 0 time 24 pv g6h7 e7e8 h7h5 e8e7 h5g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 e7c7 a5a4 f3e2 a4b3 c7d7 b3b2 d7e8 b2a2 b7b6 a2b2 e8f7 b2a2 f7f1 a2b3 f1h3 b3b2 h3h6 b2a2
info depth 17 seldepth 34 multipv 1 score cp 397 wdl 999 1 0 nodes 1537207 nps 64050291 tbhits 0 time 24 pv g6h7 e7f6 h7h6 f6e7 h6g5 e7f8 f3f4 f8e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f4f3 a7b6 f3e2 b6a5 e7d8 a5a4 d8c7 a4b3 c7d7 b3b2 b7b6 b2a2 d7e8 a2b2 e8a8 b2b3 a8h1 b3a2
info depth 18 seldepth 31 multipv 1 score cp 0 wdl 47 907 46 nodes 2879229 nps 82263685 tbhits 0 time 35 pv g6h7 e7e8 h7g6 e8e7
info depth 19 seldepth 11 multipv 1 score cp 0 wdl 47 907 46 nodes 2953521 nps 84386314 tbhits 0 time 35 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e5
info depth 20 seldepth 5 multipv 1 score cp 0 wdl 47 907 46 nodes 3018645 nps 83851250 tbhits 0 time 36 pv g6g7 e7e8 g7g6 e8e7
info depth 21 seldepth 17 multipv 1 score cp 0 wdl 47 907 46 nodes 3166946 nps 85593135 tbhits 0 time 37 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 22 seldepth 25 multipv 1 score cp 0 wdl 47 907 46 nodes 3510118 nps 90003025 tbhits 0 time 39 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 23 seldepth 25 multipv 1 score cp 0 wdl 47 907 46 nodes 3922411 nps 91218860 tbhits 0 time 43 pv g6g7 e7e8 g7e5 e8d8 e5d6 d8e8 d6e6 e8f8 e6e5 f8g8 e5g5 g8h7 g5f6 h7g8
info depth 24 seldepth 25 multipv 1 score cp 0 wdl 47 907 46 nodes 4203546 nps 93412133 tbhits 0 time 45 pv g6g7 e7e8 g7g6 e8e7 g6g5 e7d7 g5f6 d7e8 f6e6 e8f8 e6e5 f8g8 e5g5 g8f8 f3f4 f8e8 g5e5 e8f8
<snip>
info depth 47 seldepth 7 multipv 1 score cp 0 wdl 47 907 46 nodes 59505199 nps 132824104 tbhits 0 time 448 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 48 seldepth 61 multipv 1 score cp 0 wdl 47 907 46 nodes 65890441 nps 133112002 tbhits 0 time 495 pv g6g7 e7e8 g7e5 e8d8 e5h8 d8e7
info depth 49 seldepth 29 multipv 1 score cp 0 wdl 47 907 46 nodes 102910348 nps 135944977 tbhits 0 time 757 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3f4 f8g7 e6g4 g7h8 g4f5 h8g7 f5g5 g7h8 g5f6 h8g8 f6g6 g8f8
info depth 50 seldepth 8 multipv 1 score cp 0 wdl 47 907 46 nodes 103027115 nps 135919676 tbhits 0 time 758 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
bestmove g6g7 ponder e7e8

Below is the outout from functionally equivalent Stockfish XI-r3 planned to be released tomorrow, it will take into consideration, egtb, 50 move and 3-fold repetition draws.

Code: Select all

$ Sto*.exe



    ######                                                            .-'`````'-.     #     # ###             #####
    #     # #####  ####   ####  #    # ###### #  ####  #    #    (\  /   \   /   \     #   #   #      #####  #     #
    #         #   #    # #    # #   #  #      # #      #    #    ) \/   (@) (@)   \     # #    #      #    #       #
     #####    #   #    # #      ####   #####  #  ####  ######   (__(       W       )     #     #  ### #    #  #####
          #   #   #    # #      #  #   #      #      # #    #      (  V v V V v V  )    # #    #      #####        #
    #     #   #   #    # #    # #   #  #      # #    # #    #       \  A A A A A  /    #   #   #      #   #  #     #
     #####    #    ####   ####  #    # #      #  ####  #    #        \/\-.,,,.-/\/    #     # ###     #    #  #####


    Stockfish XI-r3  by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

p f 8/1p2KP2/1p4q1/1Pp5/2P5/N1Pp1k2/3P4/1N6 b - - 76 40
s t 60
info string Hash Table: Windows Large Pages, 256 Mb
Confirmation: Threads set to 60
s h 86688
info string Hash Table: Windows Large Pages, 86688 Mb
Confirmation: Hash set to 86688 Mb
g d 60
info depth 1 seldepth 2 multipv 1 score cp 44 wdl 83 904 13 sp 535 nodes 15980 nps 5326666 tbhits 0 time 3 pv g6g7
info depth 2 seldepth 2 multipv 1 score cp 26 wdl 57 924 19 sp 519 nodes 47210 nps 11802500 tbhits 0 time 4 pv g6g7 e7e8
info depth 3 seldepth 4 multipv 1 score cp 23 wdl 54 926 20 sp 517 nodes 57292 nps 14323000 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8d7 e5f5 d7e7 f5g5 e7f8
info depth 4 seldepth 5 multipv 1 score cp 25 wdl 56 924 20 sp 518 nodes 61523 nps 15380750 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8d7 e5f6
info depth 5 seldepth 6 multipv 1 score cp 21 wdl 52 927 21 sp 515 nodes 68006 nps 17001500 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8f8 e5f6 f8g8
info depth 6 seldepth 8 multipv 1 score cp 26 wdl 56 925 19 sp 518 nodes 73204 nps 18301000 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8
info depth 7 seldepth 15 multipv 1 score cp 21 wdl 51 928 21 sp 515 nodes 87340 nps 21835000 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8f8 e5f6 f8g8 f6g6 g8f8 f3f4 f8e7 g6g7
info depth 8 seldepth 14 multipv 1 score cp 9 wdl 40 932 28 sp 506 nodes 115683 nps 28920750 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3f4 f8g7 e6e7 g7g8 e7g5 g8f8
info depth 9 seldepth 16 multipv 1 score cp 35 wdl 68 916 16 sp 526 nodes 129327 nps 32331750 tbhits 0 time 4 pv g6g7 e7e8 g7e5 e8d8 e5f6 d8e8 f6e6 e8f8 f3f4 f8g7
info depth 10 seldepth 14 multipv 1 score cp 137 wdl 408 590 2 sp 703 nodes 139649 nps 27929800 tbhits 0 time 5 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3f4
info depth 11 seldepth 16 multipv 1 score cp 206 wdl 758 242 0 sp 879 nodes 153474 nps 30694800 tbhits 0 time 5 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3f4 f8g7 e6e7 g7g8
info depth 12 seldepth 22 multipv 1 score cp 226 wdl 829 171 0 sp 914 nodes 255286 nps 42547666 tbhits 0 time 6 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3e4 b5b6 e7d7 a3b5 d7c6 b1a3
info depth 13 seldepth 23 multipv 1 score cp 263 wdl 917 83 0 sp 958 nodes 358434 nps 51204857 tbhits 0 time 7 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3f4 b5b6 e7e4 a7b8 e4c6
info depth 14 seldepth 28 multipv 1 score cp 252 wdl 896 104 0 sp 948 nodes 637771 nps 70863444 tbhits 0 time 9 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3f4 b5b6 e7e4 a7b8 f4e5 a3b5 e4c4 b1a3 c4e4 b5c7 c5c4 b8a7 e4c6 a3b5 c6h6
info depth 15 seldepth 25 multipv 1 score cp 263 wdl 916 84 0 sp 958 nodes 704114 nps 70411400 tbhits 0 time 10 pv g6g5 e7d7 g5f6 d7c7 f6f7 c7b6 f7e7 b6a7 f3f4 b5b6 e7f7 a7b8 f4e5 a3b5 f7c4 b1a3 c4d5 b5c7 d5e4
info depth 16 seldepth 31 multipv 1 score cp 252 wdl 896 104 0 sp 948 nodes 1355796 nps 90386400 tbhits 0 time 15 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 f3f4 a5a4 b7b6 a4b3 f4e4 b3a2 e7e6 a2b2 e6d6 b2a1 d6f6 a1a2 f6f3 a2b3
info depth 17 seldepth 27 multipv 1 score cp 252 wdl 896 104 0 sp 948 nodes 1434676 nps 95645066 tbhits 0 time 15 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 e7d6 a5a4 f3e2 a4b3 b7b6 b3b2 d6f6 b2a2 f6e6 a2b2 e2f3 b2a2 f3e4 a2b2 e6g6 b2b3 g6f6
info depth 18 seldepth 32 multipv 1 score cp 252 wdl 896 104 0 sp 948 nodes 1486020 nps 92876250 tbhits 0 time 16 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 f3f4 a5b6 f4e5 b6a5 e7d6 a5a4 d6e6 a4b3 e5e4 b3b2 e6g4 b2a2 b7b6 a2b2 g4h5 b2b3 h5e8 b3a2 e8e5 a2b2 e5d6 b2a1
info depth 19 seldepth 32 multipv 1 score cp 252 wdl 896 104 0 sp 948 nodes 1695857 nps 99756294 tbhits 0 time 17 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7c7 f6f7 c7b6 f7e7 b6a5 e7d6 a5a4 f3f4 a4b3 f4e4 b3b2 b7b6 b2a1 d6f6 a1b2 f6e6 b2b3 e6g6 b3a2 g6g1 a2b2 g1h1 b2b3 h1d1 b3a2 d1c1 a2b3
info depth 20 seldepth 30 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 3846324 nps 113127176 tbhits 0 time 34 pv g6g5 e7e8 g5e5 e8d7 e5h8 d7e7 h8g7 e7e8
info depth 21 seldepth 7 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 3870231 nps 113830323 tbhits 0 time 34 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8
info depth 22 seldepth 7 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 3901142 nps 111461200 tbhits 0 time 35 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8
info depth 23 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 4238515 nps 114554459 tbhits 0 time 37 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8 f6e6 e8f8 e6e5 f8g8 e5g5 g8f8 f3e4 f8e8 g5g6 e8e7 g6g5
info depth 24 seldepth 11 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 4305535 nps 113303552 tbhits 0 time 38 pv g6g5 e7d7 g5f4 d7e7 f4e5 e7d7
info depth 25 seldepth 29 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 4521304 nps 113032600 tbhits 0 time 40 pv g6g7 e7e8 g7e5 e8d7 e5f5 d7e8 f5e6 e8f8 e6f6 f8g8 f6g6 g8f8
info depth 26 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 4756131 nps 116003195 tbhits 0 time 41 pv g6g7 e7e8 g7e5 e8d8 e5f6 d8e8 f6e6 e8f8 f3e4 f8g7 e6e5 g7g8 e5g5 g8h7 g5f6 h7g8 f6g6 g8f8 e4f5 f8e7 g6e6 e7f8
info depth 27 seldepth 21 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 5272370 nps 114616739 tbhits 0 time 46 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 28 seldepth 23 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 5561542 nps 115865458 tbhits 0 time 48 pv g6g7 e7e8 g7g6 e8e7
info depth 29 seldepth 10 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 5623425 nps 117154687 tbhits 0 time 48 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8
info depth 30 seldepth 11 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 5734568 nps 117032000 tbhits 0 time 49 pv g6g7 e7e8 g7g6 e8e7
info depth 31 seldepth 7 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 5897540 nps 115638039 tbhits 0 time 51 pv g6g7 e7e8 g7g6 e8e7 g6h7 e7e8
info depth 32 seldepth 39 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 6833861 nps 117825189 tbhits 0 time 58 pv g6g7 e7e8 g7e5 e8d7 e5f5 d7e7 f5h7 e7e8 h7e4 e8d8 e4h4 d8d7 h4h6 d7e7
info depth 33 seldepth 7 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 7710995 nps 118630692 tbhits 0 time 65 pv g6g7 e7e8 g7e5 e8d8 e5d6 d8e8
info depth 34 seldepth 40 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 9616931 nps 120211637 tbhits 0 time 80 pv g6g7 e7e8 g7e5 e8d8 e5f6 d8e8 f6e6 e8f8 e6e5 f8g8 e5g5 g8h7 g5f5 h7g7 f5g4 g7f6 g4h4 f6g7 h4e7 g7g8
info depth 35 seldepth 24 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 12048733 nps 121704373 tbhits 0 time 99 pv g6g7 e7e8 g7e5 e8d8 e5d6 d8e8
info depth 36 seldepth 17 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 12946179 nps 122133764 tbhits 0 time 106 pv g6g7 e7e8 g7e5 e8f8 e5h8 f8e7
info depth 37 seldepth 42 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 16689848 nps 122719470 tbhits 0 time 136 pv g6g7 e7e8 g7e5 e8d8 e5f4 d8e8
info depth 38 seldepth 42 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 19332592 nps 123926871 tbhits 0 time 156 pv g6g7 e7e8 g7e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3e4 f8g7 e6e7 g7g8 e7g5 g8h7 g5f6 h7g8 f6g6 g8f8 g6e6
info depth 39 seldepth 24 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 23700533 nps 124086560 tbhits 0 time 191 pv g6g5 e7e8 g5e5 e8d8 e5f6 d8e8 f6e6 e8f8 f3e4 f8g7 e6e7 g7g8 e7g5 g8h8 g5h6 h8g8 h6g6 g8f8
info depth 40 seldepth 16 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 30279951 nps 125642950 tbhits 0 time 241 pv g6g5 e7e8 g5e5 e8d8 e5f6 d8e8
info depth 41 seldepth 48 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 54348426 nps 126982303 tbhits 0 time 428 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8
info depth 42 seldepth 17 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 54655863 nps 127106658 tbhits 0 time 430 pv g6g5 e7e8 g5g6 e8e7
info depth 43 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 58403302 nps 126963700 tbhits 0 time 460 pv g6g5 e7e8 g5e5 e8d8 e5f6 d8e8 f6e6 e8f8 f3e4 f8g7 e6e5 g7g8 e5g5 g8h7 g5f6 h7g8
info depth 44 seldepth 24 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 61053487 nps 127194764 tbhits 0 time 480 pv g6g5 e7e8 g5e5 e8d8 e5f6 d8e8
info depth 45 seldepth 23 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 80642268 nps 127800741 tbhits 0 time 631 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8
info depth 46 seldepth 8 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 80745876 nps 127762462 tbhits 0 time 632 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8
info depth 47 seldepth 54 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 107620050 nps 128578315 tbhits 0 time 837 pv g6g5 e7e8 g5g6 e8e7
info depth 48 seldepth 19 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 121829092 nps 128783395 tbhits 0 time 946 pv g6g5 e7d7 g5f6 d7e8 f6e5 e8d8 e5g5
info depth 49 seldepth 10 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 127804888 nps 128835572 tbhits 0 time 992 pv g6g5 e7e8 g5g6 e8e7
info depth 50 seldepth 14 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 136798797 nps 128933833 hashfull 0 tbhits 0 time 1061 pv g6g5 e7e8 g5e5 e8d7 e5f4 d7e7 f4h4 e7e8 h4e4 e8d8 e4f5 d8e7
info depth 51 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 153334042 nps 129286713 hashfull 0 tbhits 0 time 1186 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8 f6e6 e8f8 e6e5 f8g8 e5g5 g8h7 g5f6 h7g8
info depth 52 seldepth 17 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 162099703 nps 129369276 hashfull 0 tbhits 0 time 1253 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8 f6e6 e8f8 e6h6 f8e8
info depth 53 seldepth 56 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 206090779 nps 129861864 hashfull 1 tbhits 0 time 1587 pv g6g5 e7e8 g5g6 e8e7 g6h7 e7e8 h7g6
info depth 54 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 260437775 nps 130545250 hashfull 1 tbhits 0 time 1995 pv g6g5 e7e8 g5e5 e8d8 e5b8 d8e7 b8c7 e7e8 c7c8 e8e7
info depth 55 seldepth 19 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 270155151 nps 130572813 hashfull 1 tbhits 0 time 2069 pv g6g5 e7d7 g5f5 d7e7
info depth 56 seldepth 19 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 297964674 nps 130686260 hashfull 1 tbhits 0 time 2280 pv g6g5 e7e8 g5e5 e8d7 e5f6 d7e8 f6e6 e8f8 f3g4 f8g7 e6e5 g7g8 e5g5 g8h8 g5d8 h8g7 d8e7
info depth 57 seldepth 16 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 307040738 nps 130711255 hashfull 2 tbhits 0 time 2349 pv g6g5 e7e8 g5g6 e8e7 g6h7 e7e8 h7e4 e8d7 e4g4 d7d8 g4f5 d8e8
info depth 58 seldepth 8 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 307327252 nps 130721927 hashfull 2 tbhits 0 time 2351 pv g6g5 e7e8 g5g6 e8e7 g6g7 e7e8 g7g6
info depth 59 seldepth 71 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 432288151 nps 131915822 hashfull 4 tbhits 0 time 3277 pv g6g5 e7e8 g5e5 e8d7 e5g7 d7e8
info depth 60 seldepth 25 multipv 1 score cp 0 wdl 0 1000 0 sp 500 nodes 503190478 nps 132209794 hashfull 4 tbhits 0 time 3806 pv g6g7 e7e8 g7e5 e8d8 e5d6 d8e8 d6e6 e8f8 f3g4 f8g7 e6e5 g7g8 e5g5 g8h7 g5f6 h7g8
bestmove g6g7 ponder e7e8

[Ovyron]

If WDL shows 200 500 300, for example, then the interpretation is that the side-to move has a 20% chance of winning and a 30% chance of losing.

That predicts that 50% of the games will be drawn. The rest are decided games. How accurate is that? Where's the data that shows when 200 500 300 is shown, that half the games are actually drawn?

Because if they are then I stand corrected and tip off my hat to WDL, as I have no way of knowing when a position has 50% chance of being decided in any way, WDL is providing incredibly useful information just like that.

But if in reality when that position is played out the decided games are nowhere near 50%, then WDL is just smoke and mirrors. People are making wrong decisions because of faulty information WDL shows them.

[zullil]

If WDL shows 200 500 300, for example, then the interpretation is that the side-to move has a 20% chance of winning and a 30% chance of losing.

That predicts that 50% of the games will be drawn. The rest are decided games. How accurate is that? Where's the data that shows when 200 500 300 is shown, that half the games are actually drawn?

Because if they are then I stand corrected and tip off my hat to WDL, as I have no way of knowing when a position has 50% chance of being decided in any way, WDL is providing incredibly useful information just like that.

But if in reality when that position is played out the decided games are nowhere near 50%, then WDL is just smoke and mirrors. People are making wrong decisions because of faulty information WDL shows them.

I think the whole thing is silly. I was just attempting to explain how to interpret the numbers. Stockfish's numbers are designed to model statistical data derived from (several million?) Stockfish self-play games (at 60s + 0.6s) on fishtest. I don't know how well the model fits the actual fishtest data. Don't care, especially given the very short time controls from which the data were generated.