- Total games starting with Tengen: 87 games, winning percentage 44%
- 1933-1936: 35 games (mostly no komi), winning percentage 35%
- 1937-1990: 14 games, winning percentage 43%
- 1991-2009: 32 games, winning percentage 56% (that is highest winning percentage with 6-4 of all opening moves in that period, though note the very limited sample)
But note two things:
- The database might be skewed towards games where tengen was a winning move, because it is more interesting and "worth writing down". When I asked Jan van der Steen, he thinks that the database is quite representative.
- There are not too many tengen games so it is maybe not statistically significant.
- Quite a number of games are from fast play tournaments. (Most notably four consecutive wins by Mok Chin-seok starting with Tengen during one particular tournament)
Herman: The latest GoGoD (games up to Dec 2009) has a total of 133 tengen games, of which black won 49.6% and white won 48.1%, the remainder left unfinished. This includes games with no komi however, as well as amateur games (you might want to check your mastergo database for this, as for example the Oteai was always played without komi, up until very recently, when it was discontinued). As for statistical significance, you can quickly estimate the margin of error based on the sample size N as 1/√N. For example with a sample size of 87 as above, the margin of error is 1/√87 which is 0.107 or 10.7%. Assuming the winning percentage of 44% is otherwise unbiased, we can say that we are 95% sure that the true winning percentage is between 33.3% and 54.7%
tapir: the younger games i have are mostly professional games with komi (with the obvious exception of the games in the 30s and oteai games which are however mostly from the 30s). i am not sure about all players of the 17th, and 19th century games, but who plays the honinbo on even seems to be professional. the main feature is imo the increasing winning percentage over the years despite introduction of komi (between between 33% and 54% is rather meaningless, isn't it?), a statistic divided according to time limits may even be more interesting. but i don't have any such information (where a tournament is given it is sometimes clear whether it is rapid or not). another interesting feature are losing strains of some players who played it more often - who subsequently stop to use it. like kubomatsu katsikuyo? four subsequent losses in 1935 after which no games are in my database for years. with some post-war tries and the first win with tengen only again in 1957.
binky? It seems there are a good many myths about the disadvantages of tengen openings. Those include:
- it is not traditional
- there is little study of it
- loss of sente
- loss of komi advantage
- no advantage for white
- there is no territory in the centre
- it leads to complex games
- don't understand it
Also i find playing white against 5 & 7 stone handicaps harder than 4 & 6 stone handicaps respectively, maybe because i don't really understand how to counter a tengen opening.
It would be interesting to see the statistics on:
- white tengen opening
- white tengen opening against 2, 3, 4, 6, & 8 stone handicap games
- 5, 7 and 9 stone handicap games (tengen given to Black)
The changes in statistics over time (decade upon decade) may reflect times when tengen opening study was popular.
Calvin: I processed the KGS high-dan game collections from 2000-2003 with Kombilo, where there where 81 tengen games. Black wins 38.3% of the time. This is compared to black winning 38.1% of the time in the entire game set of 15100 games. (The collection contains a number of even games where white is ranked higher than black, so it's important to account for this.) 81 is a small sample, and there are few players who open on tengen often, but this at least suggests that there is not an obvious decline in win rates among high-dan amateurs playing friendly games when opening on Tengen.
Dieter: Thanks for the information. Statistically, this means there is no reason to believe opening on tengen is inferior to corner play. So, while the minor differences are statistically insignificant, that very fact is quite significant.
DaveSigaty: The kicker in here is that Mok Chin-seok, one of the Korean young stars, took up tengen recently. In the GoGoD CD collection he is listed starting 5 games in 2001 with tengen and winning all 5. The two collections are a little different (the latest CD has only 20 tengen games since 1990 and 43 overall) but I think both are dominated by a few players. Yamashita Keigo is included with 8 games since 1990 on the CD, winning 5 and losing 3. The "rest of the world" did manage 4 wins to 3 losses (1 left unfinished) though.
John Fairbairn: Eh? Or have you found another bug in Kombilo, Dave? The latest edition had 83 tengen games and my own collection, which will be included in the next edition has 90. There's quite a range of young Chinese pros who play it.
- Dave I think this is just a reflection of how much work you are doing on GoGoD. I posted the numbers in May 2002. Keep it up! :-)
Andre Engels: I have some more from the 'rest of the world' - getting to 9 wins, 5 losses for the period since 1990.
- John F. Latest stats from GoGoD (summer 2003): 94 games, of which Black won 50%, White 47.9%% (not all even, of course - some at no komi). But since 1990 there have been 34 games with Black scoring 21-12 (one unfinished), AND all bar one at komi. Even the no-komi game Black won by more than enough. These games cover C, J, K and are not monopolised by any one player. In short, this reinforces the point made about Black now knowing how to use it. Incidentally, there are also games where Black plays tengen on move 3 or 5 (about another 24 in the period from 1990).