[Welcome to Sensei's Library!]

StartingPoints
ReferenceSection
About


Homepages
HolIgor

 

Holigor's Rating Of Go Players
   

The rating that is based on my own system that is very different from ELO can be found on [ext] http://msnhomepages.talkcity.com/Terminus/holigor/baduk.htm

The data on the matches are taken from Go news site at [ext] http://www.kyoto.zaq.ne.jp/momoyama/news/news.html

If you click on the names of the players you can see their record but I have to warn that the file is large and may take time to load.

The rating depends on the input information, certainly, and would change if the other tournaments were added.

The table has some additional information. You can see how well the players perform with black and white, for example.

Yi Ch'ang-ho is the leader, of coourse. Look at his record. It is impressive.

Two next places are surprising. The information on these two younsters (Gu Li and Pak Yeong-hun) is not complete though. Perhaps they do not do so well in other tournaments that I don't know about. Let's consider them new hopefuls and watch their progress.

Other remarkable features are high position of Cho Chikun despite the loss of major titles, not so good record of Ma Xiaochun and quite low position of Rui Naiwei. She has a great win/loss ratio but most wins were against other ladies and her record with men is not so great this year.

Don't pay attention to the lower part of the table. The system cannot reliably rate players with few games.


DaveSigaty: For those interested in how it works, HolIgor posted the following on RGG on Friday October 26, 2001.

barryp@es.co.nz (Barry Phease) wrote in message news:<3bd87345.61179720@News.CIS.DFN.DE>...
> On 25 Oct 2001 03:44:24 -0700, holigor@hotmail.com (holigor) wrote:
>
> >can be found at [ext] http://msnhomepages.talkcity.com/Terminus/holigor/baduk.htm%%% > >
> >The rating system is completely different from ELO.
>
> Can you elucidate?
>

I will place a description there at the nearest future. Meanwhile, I will outline it.

What I did not like in ELO is the necessity to assume an initial rating for a player. My system does not assume any, it works with the results only.

It is simple minimization of quadratic function in the form \sum_{k} a_k (x_i - x_j - r_k)^2 where the game $k$ was held between players $i$ and $j$ and produced the result $r_k$. If I add any result I just make all the calculations from the beginning. The function depends on the difference of ratings $x_i - x_j$ only and thus does not change with a shift. So I shift it after calculations to make the ratings positive. $a_k$ account for tournament based aging. For example, if the we are in the middle of meigin title match now the previous meijin title matches and leagues are taken with smaller coefficients.

After minimization one more correction is made. The ratings of people that have 1-2 games in the database are not reliable, so I reduce them artificially multiplying 1-exp (-m/5). This allows to exclude from the top players with a single win against a player in the middle of the table. This procedure practically does not effect the players with more than 20 games. Still I give raw rating in the table. You make look at who is hopeful :).

The system requires keeping all the results (at least for some time) and is hardly applicable to the large number of players but 400 players is OK. The system should work best if the scatter of the strength of the players is not too big. The function has no saturation (that's because I was can easily solve linear equations while with exponents I'd have to apply iterations. it can be done, of course). No saturation means that the stronger players are punished for playing weaklings even if they win.

> >Results of the games were taken from Go News page. They include practically
> >all major tournaments for 2 years.
>
> I wonder how you define major tournament. It looks like you include
> only international tournaments (which is fine by me), but many would
> think that the local-only tournaments are more major.
>

The results are taken from momoyama Go News page. Look there. They include international tournaments and domestic tournaments for Japan, Korea and China. Information on Chinese tournaments is not as full as for the other two.

> The results seem interesting.
>
> Why do Gu Li and Yi Sedol perform so much better with white?, most top
> players seem to get better results with black.
>
> Why is Ma Xiaochun so bad with white?

I don't know.


HolIgor: Detailed description of the rating system are at my [ext] http://msnhomepages.talkcity.com/Terminus/holigor/descript.htm

I have improved it to account for the saturation of the evaluation function for large values of rating difference. I renew it weekly.



This is a copy of the living page "Holigor's Rating Of Go Players" at Sensei's Library.
(C) the Authors, published under the OpenContent License V1.0.