>From CGOS data it looks like WHITE has the better winning chances with komi set at 7.5.
Take this all with a grain of salt because data like this can be misleading. The weaker moves, for instance, may be better than they appear due to the possibility that the poorer results are caused by weaker players using these moves. Or some moves may be good but very difficult to play correctly. I processed the data using various criteria, right now I'm looking at games where BOTH sides are at least 1700 CGOS rated and both players are within 100 ELO of each other. Here is how the game should be played statistically: If you are the first player, you can play any of the 8 points surrounding the center or the center itself. These are 3 canonically identical moves and if you mini-max whites responses they are all approximately equal in value. For instance E5 is stronger overall, but if you consider the opponents specific responses, white does well against E5 too. So here are the "recommendations" the CGOS database: Against E5, the strongest reply seems to be C5, giving white a pretty good edge of 56.0% Against D5, white should play F5 and has a 56.1% advantage. Against D4, the data is not quite as clear. The strongest response is C3 with a whopping 61.7% win rate, but based on only 141 samples. The most popular response is E6 but this only gives 51.4% out of 1466 games. The second most popular response is E5 and it seems to give better results, scoring 55.2% out of 1355 games. Beyond this, other first moves quickly degrade in effectiveness. For instance C4 is punished with several good choices around the center, E5 being the top choice and wins 60.1% out of 454 games. I am showing below an abbreviated chart. I'm taking out any responses without reasonable samples or scores for brevity and remember that all scores are from the point of view of the players who played the move in quiestion. E5 49.1% 19630 | C4 49.6% 5894 | C5 56.0% 4131 | C3 54.2% 3702 | D4 50.5% 3634 | D5 39.3% 1688 | B4 46.9% 258 | B5 43.7% 206 D5 47.9% 8404 | F5 56.1% 2465 | F4 52.9% 1516 | E5 50.8% 758 | G4 51.9% 697 | E4 49.1% 696 | F3 49.5% 553 | G5 52.9% 420 | D3 56.7% 270 | C4 48.8% 201 | E3 46.3% 177 | G3 40.0% 145 | C3 46.8% 126 | D4 41.3% 126 D4 47.4% 7963 | E6 51.4% 1466 | E5 55.2% 1355 | F6 53.1% 1019 | F7 51.3% 967 | D6 58.7% 652 | G7 58.5% 535 | C6 50.5% 475 | E7 49.7% 352 | D7 56.1% 330 | C7 40.4% 213 | D5 47.0% 202 | C3 61.7% 141 C4 44.6% 3206 | E5 60.1% 454 | F6 58.3% 422 | D5 59.6% 317 | F5 57.7% 246 | G6 54.3% 184 | E4 56.3% 174 | F3 49.7% 171 | E6 54.2% 153 | F4 58.7% 150 | G7 48.6% 142 | E3 58.1% 105 | G3 50.5% 101 C5 47.9% 2270 | E5 47.9% 453 | F5 57.3% 426 | F4 50.3% 322 | G4 64.4% 267 | F3 43.6% 266 | E4 50.8% 187 | D4 57.1% 105 | D5 49.1% 53 C3 40.3% 1438 | E6 56.1% 173 | E5 52.1% 169 | F7 68.3% 161 | D4 66.7% 150 | F6 67.1% 149 | G7 57.5% 120 | D5 58.0% 100 | D6 57.7% 97 | C6 60.2% 93 | D7 71.8% 39 | E7 60.0% 35 _______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
