Best Players in the World

[emeraldemon]

Another thread (http://www.lifein19x19.com/forum/viewto ... =13&t=1365) brought up a question: who are the best players in the world right now? Of course, it's not possible to answer definitely, but we can try anyway

Here's my top 5:

1. Lee Sedol
2. Lee Changho
3. Gu Li
4. Kong Jie
5. Chang Hao

Here's my top 24:

1. Lee Sedol (30)
2. Lee Changho (29)
3. Gu Li (25)
4. Kong Jie (20)
4. Chang Hao (20)
6. Park Yeonghun (19)
7. Hu Yaoyu (10)
8. Piao Wenyao (9)
8. Choi Cheolhan (9)
10. Huang Yizhong (6)
10. Qiu Jun (6)
10. Cho Hanseung (6)
13. Kang Dongyun (5)
13. Xie He (5)
15. Zhou Junxun (4)
15. Han Sanghoon (4)
15. Liu Xing (4)
15. Zhou Ruiyang (4)
15. Zhou Heyang (4)
20. Kono Rin (3)
20. Yamashita Keigo (3)
20. Cho U (3)
20. Wang Xi (3)
20. Yoda Norimoto (3)

I made that list by collecting results from the major international tournaments in the last three years (thanks to Mr Kin's Go News). I gave 4 points for a win, 3 for challenging, 2 for making top 4, and 1 for making top 8. Points are in parenthesis after the name. Of course there are arguments about how these international tournaments don't really represent the best players in the world, but I don't really find those arguments convincing.

So, what's wrong with my list? Who do you think are the best?

[SoDesuNe]

I'd take a general win/loss-ratio of all games played and by that (as John Fairbairn pointed out somewhere) Yi Chang'ho is pretty much outstanding.

[hyperpape]

A four way tie for best in Japan?

No, this is not right. See http://senseis.xmp.net/?ChoU, if you're confused.

[emeraldemon]

I'd take a general win/loss-ratio of all games played and by that (as John Fairbairn pointed out somewhere) Yi Chang'ho is pretty much outstanding.

That's an interesting idea. I have GoGoD, although it's a few years old. Is there a way to do this with Kombilo maybe?

A four way tie for best in Japan?

No, this is not right. See http://senseis.xmp.net/?ChoU, if you're confused.

I don't see anything on that page that I missed. Here are the results I have for those four from Japan:
Cho U:
12th LG Cup (T8)
20th Fujitsu Cup (T4)

Kono Rin:
22nd Fujitsu Cup (T8)
13th LG Cup (T8)
12th LG Cup (T8)

Yamashita Keigo:
22nd Fujitsu Cup (T8)
13th LG Cup (T8)
13th Samsung Cup (T8)


Yoda Norimoto:
21st Fujitsu Cup (T8)
20th Fujitsu Cup (T4)

Remember I was only including results from the last three years, so I didn't count the 9th LG Cup win for Cho U. You could expand the window out to 5 or 10 years and see how it changes, but that would be more work . I'm not even sure how I would get those results, since Mr. Kin's doesn't go back that far. Maybe I could find some clever way to parse the sgfs in GoGoD looking for tournament results, but that also sounds like work.

[hyperpape]

Yes, they may be equivalent in international competition in the past three years, but I think it's irrelevant, because Cho dominates each of those players in Japanese competition, and there's far more of that. He's simply at a different level than any other Japanese professional, except perhaps Iyama Yuta.

Actually, I wouldn't put Kono or Yoda in the cohort immediately behind Cho, at least not if we are thinking of the period since 2005. For the Japanese, I think the top two tiers are probably five players:

Cho, Iyama, Yamashita, Takao, Hane (the last three are not necessarily in any order).

More generally, I think it doesn't make sense to ignore each nation's tournaments. International titles are very prestigious, but if a player has a lot of wins in his nation's competition, that tends to suggest he's better than his countrymen. That's not to deny that international titles are more impressive wins than single-nation titles, but ignoring one gives you perverse rankings.

[emeraldemon]

...
For the Japanese, I think the top two tiers are probably five players:

Cho, Iyama, Yamashita, Takao, Hane (the last three are not necessarily in any order).

More generally, I think it doesn't make sense to ignore each nation's tournaments. International titles are very prestigious, but if a player has a lot of wins in his nation's competition, that tends to suggest he's better than his countrymen. That's not to deny that international titles are more impressive wins than single-nation titles, but ignoring one gives you perverse rankings.

So do you have a top 5 or top 10 worldwide? I can understand taking national tournaments to compare players within those countries, but I hesitate to use those results to move someone up or down the ranking. Maybe some sort of elo-type approach, estimating a "real rank" of the players?

[freegame]

nr 1:

Fujiwara Sai

1. Lee Sedol
2. Lee Changho
3. Gu Li

this looks like a nice top 2-4.

[hyperpape]

I don't know a ton about rating systems, but ELO style seems most reasonable, especially as you have bigger pools. A system based on places in major tournaments often works well for the very top players, not so well for players who might have a single place, which can be subject to a lot of chance variation.

[Magicwand]

best player i think is
l. lee sehdol
2. kong jae
3. guli

lee changho is not strong as before. but i think he can be #4...
cho-u? how can he be in top 20??

[John Fairbairn]

Maybe I could find some clever way to parse the sgfs in GoGoD looking for tournament results, but that also sounds like work.

It's very easy if you use GoLibrary on the disc (Windows only, of course). You can, for example, first search on a tournament, or even a round, then search further on a player. You can do the player search on Black, White or both. You can limit the year range. There are other filters. Win-loss totals and ratios come up automatically. I expect several other programs such as Drago allow similar and better procedures. Remember that we don't hide GoGoD behind cyphers or passwords, so that quite a few people have made use of the data in creative ways. Outside the 64,000 sgf files, GoGoD will also give you much data, background or gossip on the tournaments, as well as biographical data on, currently, 3,900 people.

Yes, it sounds like work. But we've already done most of it!

[hyperpape]

cho-u? how can he be in top 20??

No clue: http://igo-kisen.hp.infoseek.co.jp/world06.html; http://senseis.xmp.net/?LGCup

You're probably not aware, but he hasn't been playing much in international titles. He's missed the 12th-14th samsungs, as well as the 14th and 15th LG cups (that's not an exhaustive list).

[xed_over]

It's very easy if ... (Windows only, of course).

one of these days I'm gonna have to try that Windows thing

[hyperpape]

John: can you do a search for games by a player just in international competition?

[John Fairbairn]

John: can you do a search for games by a player just in international competition?

Yes, in the sense that you can first search for Samsung, and then add the results to that a search for LG Cup, and so on.

As a little test, I searched for Samsung, LG, Fujitsu, Toyota, Ing, Chunlan and Nongshim in just a few seconds. I got about 4,000 hits, so there's probably enough data to give some interesting insights. Not that I'm up for it, mind!

[hyperpape]

I don't know why I even ask: I already know I'm putting GoGoD on my christmas list.