I have only just been introduced to marumaru0413's mammoth ratings site by one of my friends. Unfortunately it is all in Japanese but it encompasses Japanese and international professionals in a sprawling work-in-progress that first appeared online in 2008. Since 2011 the ratings have been based on glicko-2, one of the well-known rating systems originally developed for chess. The two main ratings lists, one for Japanese pros and the other for CJK (including Taiwan) pros, are updated monthly. The material is free to reprint and use, with no warranty from marumaru0413 ["リンクや内容の転載については自由にしてください．その場合に発生する問題(内容の間違いなど)については責任を持ちません．"]. So far I have not found any explanation of who mamumamu0413 is.
Among the features:
- The international professional listing that rates and ranks professionals that have an RD of 0.5 or less (926 in October 2014). RD stands for ratings dispersion, which is a measure of the uncertainty in the rating as calculated by the glicko-2 method. A smaller RD indicates a more reliable rating. So all pros with a reasonably reliable rating.
- The Japanese professional listing that rates and ranks all Japanese professionals with a stable RD (including some over 0.5 that have been at or below 0.5 in the past). It covers 426 pros in October. The source data and rating results are different between the two lists (the J-pro list is based only on domestic tournaments). So people appearing in both lists have slightly different ratings.
- The top ten pros on the J-pro list for every month since January 1951! Go Seigen was number one in January 1951 (actually for the first two years and nine months). On the other hand, Cho Chikun has headed the list more often than anyone else, 146 months (followed by Sakata's 133 months).
- A lengthy analysis of the relative strength of all the characters in Hikaru no Go, including a timeline of their development during the series!
- An analysis of the relative strength of insei and of all the people who qualified as professionals since 2002.
And much more!
Here is a graph I made from mamumamu0413's J-pro list. It shows the extreme dispersion of ratings by dan. For each dan level (9 through 1) the graph shows the maximum, mean, and minimum rating among all players at that rank. Note that a ratings difference of 1.0 means that when two players meet, the higher rated player has a 75% chance of winning versus 25% for the lower rated player. A ratings difference of 2.0 means that the higher rated player has a 90% chance of winning. IMHO the graph clearly shows how meaningless dan ranking has become as a measure of playing strength.
To put it in perspective here is a table showing the probability of the average X-dan in Japan defeating 16-year-old Kyo Kagen, 2p in an even game. Why did I use KK? Because he has achieved the highest ever first-year rating of any new pro in Japan (according to the newcomer analysis noted above) and is currently ranked 12th overall in Japan by rating.
