Iyama's world ranking

[hyperpape]

Can anyone who knows Mamumanu’s system comment on how much weaker it thinks Iyama is today than at his peak? It seems like goratings has drastically downgraded him (from a probably inflated peak), while Mamumanu has made a smaller adjustment (from a more accurate starting point).

[TheCannyOnion]

More interesting is that Ichiriki Ryo is at 36th, two positions above him.

Inflated! Goratings cannot be trusted when it comes to Japanese players. Mamumamu ranks Ichiriki at 75th, still 15 places behind Iyama.

[jlt]

Here are Iyama's and Ichiriki's rankings according to jlt's ranking system described a few messages above (assuming I didn't misclick, as I did a lot of manipulations by hand on Openoffice).

1) If we only take into account the most recent matches against Chinese and Korean opponents during a period of 1 year, then Iyama gets an Elo rating of 3409 (56th) and Ichiriki's rating is 3410 (55th).

2) Same thing for 2 years instead of 1 year: Iyama's rating is 3421 (47th) and Ichiriki's is 3343 (93th).

I wouldn't say any of these rankings is reliable since they fluctuate too much, but they do reflect the fact that Iyama's level has decreased while Ichiriki is catching up.

The calculation (1) doesn't take into account Iyama's 4 recent losses (3 against Murakawa and 1 against Ichiriki).

[TheCannyOnion]

Iyama has dropped to #63 on Mamumamu's ranking. On Goratings.org, which greatly inflates Japanese players' strength, Iyamma has dropped to #42.

#42 is still too high, but that's Goratings for you.

[Elom0]

I'll take the opportunity to post this while by78 and TheCannyOnion are not active at the moment *undercover movements*.

I think a main reason why Iyama Yuuta was overrated was because Goratings uses normal distribution. Indeed we often see this with #1 and #2 player, well at least I do. The rating difference between the top two players in a normal distribution almost always seem ridiculously large to me. So we have a normal distribution. Then we have almost a separate player pool in the case of Japan, and a pool highly represented in the case of the Women's Korean Baduk League. This results in Iyama Yuuta being overrated and Cho'ie Jeoung's rating being much higher at the end of the Women's Korean Baduk League. Of course the player pool skew is going to be an issue for most rating systems but . . . they can easily be addressed by reducing the weighting of games between players who play in in the same country, and of female-female games, both to about half of the standard. If this would reduce the overall weighting of games too much, instead one can multiply the standard rating by root2, and the same-country and female-female pairings by root1/2, to get the same 2 to 1 ratio but with an average of 1. I tried desperately to contact Rémi Coulom about changing the weights in this way, but I wasn't successful. Might anybody know how to reach him with this message?

I intend to create a new rating system based on an extremely simple version of WHR, using logistic distribution with games from international tournaments and between players who, relative to that point in time, rarely play each other or play against players who play each other, all by the end of this year, but I have a reverse-talent for programming. To me the 15-year-old hackers and programmers are a far more inexplicable existence to me than 15-year-olds doing last-year undergraduate calculus.

[xiaodai]

On this ranking, Iyama Yuta is currently 10th in the world which is quite exciting https://daizj.net/baduk-go-weiqi-ratings/