Sumo ratings from May?

[Sesostris 0]

Hello all.

I notice that Doitsuyama's sumo rankings page hasn't been updated with the results of the May Basho. Will the site no longer be updated, or do I just have to be patient?

The reason for my anticipation is that, as far as I can see, Asashoryu has a rating of 2580 after Haru 2005, and that's a tie with Takanohana for highest score ever. So naturally, I'm curious as to how high Asa will go after a zensho yusho...

PS: Does a wrestler get a bonus in the ratings for going zensho? After all, his strength for that yusho is effectively immeasurable.

[Doitsuyama 1,259]

Just be patient... there is no special bonus for a zensho yusho. The ratings are done on a bout-per-bout base anyway, not per basho. Fusen bouts aren't included, kettei-sen are included.

[Sesostris 0]

Thanks for the answers. (In a state of confusion...)

Just one little nitpick: Is Kyokutenho really the highest rated wrestler? (Showing respect...)

[Guest Vincentoryu]

Does a wrestler get a bonus in the ratings for going zensho? After all, his strength for that yusho is effectively immeasurable.

His basho performance (tournament performance in chess) is effectively immeasurable but his overall rating can easily be calculated using the formulas of Doitsuyama.

As Doitsuyama said he calculates the ratings on a bout for bout basis. In chess all games played in a tournament are calculated with the pre tournament ratings of the players so the first game is as important as the last game for calculating the new ratings.

Great work!! (In a state of confusion...)

[Doitsuyama 1,259]

Yes, right. In chess (i.e. tournament based rating or more correctly, half-year based rating) each game in the rating period is of same importance. But I think it is slightly more correct to have the ratings match-by-match, so the last bout is slightly more important for the rating than the one before and so on (weights degressing exponentially).

@Sesostris: Thanks for the pointer. It's corrected now.

[Zuikakuyama 1]

I am not a ratings statistician, but I do play some chess, so I am vaguely familiar with how the chess ratings work.

I know that in chess, it gets progressively harder to change your rating as you move up the rankings. Is there a similar system you are using. As far as I can tell from the explanations on your site, the only thing that slows down rating changes are the number of bouts a rikishi has had.

I think that in chess ratings, if you beat someone 400 points or lower than you are, you get no point increase for the win since you are expected to win all matches. What is that point with your system?

[Doitsuyama 1,259]

I am not a ratings statistician, but I do play some chess, so I am vaguely familiar with how the chess ratings work.

I know that in chess, it gets progressively harder to change your rating as you move up the rankings.  Is there a similar system you are using.  As far as I can tell from the explanations on your site, the only thing that slows down rating changes are the number of bouts a rikishi has had. 

The thing in chess is done to adjust between amateurs and professionals. We are speaking about rating differences between 1400 (or less) and 2800 in chess. My Sumo ratings are solely for Juryo and Makuuchi. I don't see a need to adjust rating changes depending on the level of the rating.

I think that in chess ratings, if you beat someone 400 points or lower than you are, you get no point increase for the win since you are expected to win all matches.  What is that point with your system?

Are you sure with the chess thing? This sounds like an adjustment per match which is unusual in first place. Anyway, "you are expected to win all matches" is a bit stupid, more precise is about 92% with 400 points difference.

Nowadays Asashoryu can have a more than 400 points higher rating than his aite, like against Kotooshu (401), Iwakiyama (439) and Kaiho (494). And even in the case of Asashoryu against Kaiho I wouldn't bet on a win for Asashoryu if the odds are really short, like 100-1 for Asashoryu (you bet 100 and win 1). So a small increase in rating is justified. Asashoryu won and his rating increased by 0.84 points. A loss would have lowered his rating by 19.16 points.

[Sesostris 0]

Thanks for all the info guys.

But I still have my original question: If I'm not mistaken, Asashoryu now has the highest ranking ever since the start of Doitsuyama's ratings table (1989). Since the ratings should also take the level of the opposition into account, does that make him 'best', or does it mean the system should be adjusted, or what?

There have been many discussions on this board about Asa vs. Taka vs. Chionofuji vs. Taiho, and with his current rating, his peak form seems to lie above that of Taka, at any rate. The question is how long he can maintain his present form, but in the hypothetical Asa-at-his-peak vs. Taka-at-at-his-peak, this seems to suggest that Asa should win.

[Randomitsuki 2,898]

Thanks for all the info guys.

But I still have my original question: If I'm not mistaken, Asashoryu now has the highest ranking ever since the start of Doitsuyama's ratings table (1989). Since the ratings should also take the level of the opposition into account, does that make him 'best', or does it mean the system should be adjusted, or what?

<{POST_SNAPBACK}>

First of all, I love these Elo-based rankings Doitsuyama provides. However, although I believe they are the best indicator of strength, it is difficult to compare Asashoryu's current rating with that of former yokozuna.

To illustrate this example, take one of my favorite sites for comparing the strengths of national football teams, to be found here.

They are using a slightly adapted version of the chess-based Elo ratings or Doitsuyama's ratings. The problem with all these rating systems are relatively isolated entries. For instance, the football team of Australia is notoriously and literally over-rated in those ratings because Australia plays 90% of matches against other teams from Oceania, and they win almost all of these games. In terms of Elo ratings this means they get more and more points because they do not face strong opposition.

Applied to sumo this means that if Asashoryu does not face strong opposition and continues with zensho or near-zensho records, his ratings will go continually upwards (though only by small increments). And still, I believe it is impossible to say based on these ratings that he is better than (insert favorite retired yokozuna) because those rikishi of old might have had stronger opposition.

If Asashoryu has the highest rating ever recorded by Doitsuyama we can only infer that he is more dominant than earlier rikishi. No big surprise, if you ask me.

Edited June 22, 2005 by Randomitsuki

[Doitsuyama 1,259]

In principle the sumo ratings don't have the problem of the mentioned football ratings. Oceania is a pretty secluded group not playing very often against other teams which makes it a rating group for itself, so ratings develop which can't be compared.

In Ozumo basically everyone fights everyone, so noone or no group is secluded, so the ratings are pretty much ok from this base.

A bigger problem is comparing ratings from different eras which can't really be solved by any system. I still think that Takanohana and Asashoryu are comparable enough. They aren't too far separated and even have many opponents in common. So, yes, I'd say ratingswise Asashoryu's peak is higher now than Takanohana's peak. Considering the strength of opponents, don't forget that Takanohana had a LOT of easy opponents in 1996 due to heya-advantage and a general low strengthwise.

I wrote an article earlier comparing ratings back to 1973 (now where is that...) and you will easily see that Asashoryu still lacks a lot until reaching Kitanoumi. Now that I have all Makuuchi bouts back to 1926 I should rewrite that article covering a bigger timeframe. Honestly comparing ratings will be almost non-possible then, especially for the years before the six-basho-per-year era.

[Shimpu 0]

There have been many discussions on this board about Asa vs. Taka vs. Chionofuji vs. Taiho, and with his current rating, his peak form seems to lie above that of Taka, at any rate. The question is how long he can maintain his present form, but in the hypothetical Asa-at-his-peak vs. Taka-at-at-his-peak, this seems to suggest that Asa should win.

I don't agree... Asa fought twice against Takanohana and lost. They ware in the similar shape. So, I think this ratings gives some fun but is is only fun... Taka won against Asa and this is a fact...

[Randomitsuki 2,898]

I don't agree... Asa fought twice against Takanohana and lost. They ware in the similar shape. So, I think this ratings gives some fun but is is only fun... Taka won against Asa and this is a fact...

<{POST_SNAPBACK}>

That fact is beside the point because when Asashoryu met Takanohana in Aki 2002 his rating was close to 2400 which is about the rating strength that Kaio and Tochiazuma currently have. Although Takanohana surely was declining back then his rating was much better (around 2520). The question that I recall was about Taka-at-peak (way before their bouts) and Asa-at-peak (now or later).

Edited June 22, 2005 by Randomitsuki

[aderechelsea 125]

I don't agree... Asa fought twice against Takanohana and lost. They ware in the similar shape. So, I think this ratings gives some fun but is is only fun... Taka won against Asa and this is a fact...

<{POST_SNAPBACK}>

You fail to understand that Doitsu spoke about both of them being at their peak.

I don't think that you consider Asashoryu's peak in 2002 (Eh?)

[Shimpu 0]

don't think that you consider Asashoryu's peak in 2002 huh.gif

Yes, but that was also end of Takanohana. He was much weaker then in 90's. I think Asa was in shape enough too compare him with Taka out of shape (Bleh!) ... But, it is not a point of Sumo ratings... as you've notice...

[Doitsuyama 1,259]

don't think that you consider Asashoryu's peak in 2002 huh.gif

Yes, but that was also end of Takanohana. He was much weaker then in 90's. I think Asa was in shape enough too compare him with Taka out of shape (Bleh!) ... But, it is not a point of Sumo ratings... as you've notice...

<{POST_SNAPBACK}>

Now things are really getting out of whack. Hey, in those two basho Takanohana had just a memorable 13-2 kettei-sen yusho after a 13-0 start and blowing out his knee against loose-mawashi Musoyama on day 14 and a 12-3 jun-yusho after an ultimate senshuraku showdown loss against Musashimaru. Out of shape? You are kidding. Asashoryu meanwhile was Komusubi 8-7 and Ozeki 10-5 in these two basho, not exactly his peak. Those two bouts aren't representative for equal match-ups. The ratings were 2150 vs. 2505 and 2424 vs. 2509.

Sorry, but: did you even view these two basho?

[Shimpu 0]

The ratings were 2150 vs. 2505 and 2424 vs. 2509.

Yap... it leads nowhere...

I don't like numbers - for me it isn't enough to illustrate someone skills.

What about Chiyo - Taka duel ? Chiyonofuji was on the top Sumo rating... but Taka was able to beat him... counter to digits.

Yes, it was't Asas peak but it was't Takas peak too. Ok. I'm fading away... I'm not going to bend yours ear (Eh?)

[Zentoryu 154]

What about Chiyo - Taka duel ? Chiyonofuji was on the top Sumo rating... but Taka was able to beat him... counter to digits.

<{POST_SNAPBACK}>

Just because rikishi A has a higher rating than rikishi B doesn't mean rikishi A is going to win every single time he faces rikishi B. It just indicates that Rikishi A has a better percentage chance of winning.

So it really doesn't make any sense to point at one fight where the rikishi with the higher rating lost and use it as an example that the system is flawed.

And like it or not, these ratings have been fairly accurate over the years in predicting bout results, and I suspect is one of the reasons Doitsuyama is one of the best sumo games players out there.

Edited June 23, 2005 by Zentoryu

[Doitsuyama 1,259]

The ratings were 2150 vs. 2505 and 2424 vs. 2509.

Yap... it leads nowhere...

I don't like numbers - for me it isn't enough to illustrate someone skills.

Then what the heck are you looking for in this thread? It's really only suitable for people who like numbers.

[Asashosakari 20,225]

I wrote an article earlier comparing ratings back to 1973 (now where is that...)

This, presumably?

and you will easily see that Asashoryu still lacks a lot until reaching Kitanoumi. Now that I have all Makuuchi bouts back to 1926 I should rewrite that article covering a bigger timeframe. Honestly comparing ratings will be almost non-possible then, especially for the years before the six-basho-per-year era.

<{POST_SNAPBACK}>

In case you need any encouragement for what sounds like a fairly big endeavour, I'd sure like be interested to see those numbers. (Eh?)

[Jejima 1,488]

Where can Doitsuyama's rating page be found exactly, please?

I've just checked the web-page for football mentioned by Randomitsuki. I don't think that Australia are *too* out of place at #30. Remember the other big power house of Oceania (New Zealand who only ever lose to Oz in those regional matches) are at 75 also don't seem to out of place compared to the other teams nearby.

I'm surprised to see England ranked above Brazil though....

[Randomitsuki 2,898]

I've just checked the web-page for football mentioned by Randomitsuki.  I don't think that Australia are *too* out of place at #30. 

<{POST_SNAPBACK}>

[off topic]

Yes, but now is one of the rare occasions where Australia faces international opponents (Confederations Cup). Before the Cup started they were ranked 19, along with teams like Denmark or Colombia which is probably too high.

[\off topic]