search

zxkyjimmy / ewin-data

I would release the match information for the exact-win strategy here.
1 stars 0 forks source link

could you explain the meaning of winrate in the sgf files #1

Open l1t1 opened 5 years ago

l1t1 commented 5 years ago 
(;GM[1]FF[4]RU[Chinese]DT[20180914-180436]SZ[19]KM[7.5]TM[3600]PB[Leela Zero]PW[Exact win]RE[W+Resign]C[W+Resign]
;B[pd]C[45.82%];W[dd]C[54.06%];B[pp]C[46.04%];W[dq]C[54.07%];B[co]C[46.03%];W[cp]C[54.05%];B[do]C[45.96%];

B[pd]C[45.82%]is this the winrate of black calulated by black (Leela Zero)?

zxkyjimmy commented 5 years ago

For example, in the file 001.sgf,

(;GM[1]FF[4]RU[Chinese]DT[20180914-180436]SZ[19]KM[7.5]TM[3600]PB[Leela Zero]PW[Exact win]RE[W+Resign]C[W+Resign]
;B[pd]C[45.82%];W[dd]C[54.06%];B[pp]C[46.04%];W[dq]C[54.07%];B[co]C[46.03%];W[cp]C[54.05%];B[do]C[45.96%];

First move is B[pd]C[45.82%], and the log 001-leela.log said

 Q16 ->    7902 (V: 45.82%) (N: 21.26%) PV: Q16 Q4 D16 C4 E3...

Second move is W[dd]C[54.06%], and the log 001-exact.log said

 D16 ->    8935 (V: 54.06%) (N: 38.22%) PV: D16 Q4 D3 C5 C4 D5...

And so on.

l1t1 commented 5 years ago

i see. the 2nd move of 001-leela.log is as following, is the different due to ewin method?

  Q4 ->    3041 (V: 54.15%) (N: 38.17%) PV: Q4 D16 C4 E3 D3 E4 C6 O3 R6 Q3 R3 R2 P3 Q2 P4 R4 P2 R5 Q6 K3 R14 R12 P14 O16 P11 S6 S7 S5 P8 S16 C17
 D16 ->    2932 (V: 54.11%) (N: 38.55%) PV: D16 Q4 D3 C5 C4 D5 F3 C14 F17 C16 C17 B17 C15 B16 D15 D17 B15 E17 F16 C10 O17 F18 R17 Q17 R16 Q15 Q18 P18 R18 G18 C18 B18 E16 D18 C12 D12
  D4 ->    1930 (V: 54.35%) (N: 18.69%) PV: D4 Q4 D17 C15 C16 D15 F17 C6 F3 C4 C3 B3 C5 B4 D5 D3 B5 E3 F4 C10 O3 F2 R3 Q3 R4 Q5 Q2 P2 R2 G2 C2 B2