Background Noise

I have posted the summary results from the Japanese pro database before. Here is a little more detail.

1. Games with no komi in the Oteai, Saiko-i and Saikyo tournaments There were 30,876 games. Black won 17,519 (58.1%) and White 12,626 (41.9%). 731 were ties or no results.
2. Games with 4.5 or 5.5 komi in the period 1960-75. Regrettably they do not distinguish games according to komi, but the following were with 4.5 at least most of the time:
   • Honinbo
   • Nihon Kiin Ch'ship
   • All Japan 1st Place
   • Nihon Kiin 1st Place
   • Hayago Meijin
   • Young Players
   The following were mostly 5.5:
   • Meijin
   • 10-dan
   • Oza
   • Prime Minister's Cup
   • NHK
   • Hayago
   • New Stars
   • New Faces
   • Japan Series
   • Pro Best 10
   • Chubu Top Place
   Tokyo Shinbun Cup was 5.0 and play again if jigo. There were a few other tournaments I haven't checked. Allowing for the fact that certain tournaments have far more games than others, I'd estimate the split as roughly 40:60 for 4.5 against 5.5 or, if anything, slightly less. There were 19,571 games in this category. Black won 10,402 or 54.5% and White won 8,691 (45.5%), with 478 tied/no result.
3. Games with 5.5 komi since 1975 (when 4.5 was abolished) There were 51,711 games. Black won 26,882 (53.0%) and White won 23,798 (47.0%). 1,031 were ties/no result (these are excluded from all percentages here).

To repeat for ease of comparison:

The size of the komi therefore appears to have a clear effect: at first glance 1 point komi is not too far off 1 percentage point. However, there is a curious phenomenon to account for. In the period 1986-95 (i.e. roughly half the period of 5.5 komi alone) White improved his ratio slightly (Black won 16,146 games = 52.7%, White 14,510 ), which implies it was worse in the first half of this period. This is attributed by the Nihon Kiin to the beneficial effect (for White) of nirensei by White, popular in that period, and the relative failure of the sanrensei for Black (I have already posted these figures).

Another oddity is that Black does even better in 7-game title matches. In these he has a winning ratio of 54.2%. Possibly the very best players can exploit Black's inherent advantage a little bit more? Tentative proof of this is a comparison of the winning ratios for the top players, viz.

Two players in this list (O Rissei and Ryu Shikun) actually have a higher winning ratio with White, but in both cases they are well above 70% with either colour. There are several more oddities lower down the lists. E.g. Enda Hideki has a 64.6% winning ratio with White but only 51.5% with Black - the biggest discrepancy this way round. There are lots of players who have big winning ratios with Black and small ones with White. One big gap is Kanashima Tadashi 9d (Black 67.3%, White 49.0%). All these figures are based only on komi games for players who have played at least 200 games.

-- John Fairbairn