Jump to content

 

Photo

Somewhat Historical Rikishi Strength Analysis


  • Please log in to reply
32 replies to this topic

#26 Doitsuyama

Doitsuyama

    Yokozuna

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 7,512 posts

Posted 26 January 2005 - 08:50

The case of Kaio has always puzzled me.

He is not like other rikishis whose strength deteriorates over time.  Rather his rating has gone up (from 2300 to 2400) in the last 5-10 years, despite having numerous well known injuries.  Is he qualitively better now than during the Ake-Taka era?

Is the system fully accounting for the strength of the opposition?

As a comparision, Musoyama has also gone up since year 2000 when the 2 yokozunas were not active.  Musashimaru has experienced less fluctuations, but there is still a slight increase.

<{POST_SNAPBACK}>

I would say, yes, of course Kaio got stronger over years. Just look at his failed Ozeki runs, and the time of the yusho. It's pretty obvious to me that he got better over the years, and the ratings just document that.

And yes, the system is specifically designed to account for the (rated) strength of the opposition. I write rated strength, because the actual strength (influenced by injuries, development curve etc.) is next to impossible to get factured in.

#27 MongolPower

MongolPower

    Makushita

  • Inactive Members
  • PipPipPipPip
  • 194 posts

Posted 26 January 2005 - 10:25

Thank you very much Doitsuyama!

There were criticism at the time of yokozuna promotion of Asashoryu that he was doing well because the competition was low quality. But many of current rikishi are rated quite high considering list contains best of 22 years.
Is it current competition is low in quality and Asashoryu is doing well
OR
Asashoryu is doing well so competition seems low in quality?
Respect and be Respected!

#28 Doitsuyama

Doitsuyama

    Yokozuna

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 7,512 posts

Posted 26 January 2005 - 11:28

As I have stated already, my opinion is that the competition today is not weak. The "problem" at the current time is that there is no rikishi at hand who would really challenge Asashoryu for the yusho. Only if Asashoryu has an off basho, someone else wins the yusho, and then it is up for grabs with no clear favorite.

Kitanoumi certainly didn't have this luxury which is what limited his overall yusho count. I think Kitanoumi's case was the unusual one, not Asashoryu's. Most strong Yokozuna have a period of varying length without real yusho contenders, Chiyonofuji, Takanohana and Musashimaru (at the tail end of his career) or even Akebono (at the beginning of his Yokozunadom, but even then Takanohana plagued him already).

So I think Asashoryu should just feel lucky as long as he can, rack up all the yusho, and don't let critics questioning his quality come to his heart (not that I think this would happen anyway).

#29 Kashunowaka

Kashunowaka

    Ozeki

  • Archived Members
  • PipPipPipPipPipPipPipPipPip
  • 2,167 posts

Posted 26 January 2005 - 14:07

So, what can we conclude from last bashos results and ratings? I picked out some results which I found interesting. The ELO rating system in itself isn't flawless, but it often does a pretty good job, as you will see:

Asashoryu (15-0 / expected wins 12.6) cannot do more than this to improve his rating. And he didn't get the chance to beat number 2 (Kaio), which would have improved his ratings even more. This is one possible flaw with the strength rating: your true strength can't be measured if you win all matches. (You are going off-topic...)

Chiyotaikai (8-7 / expected 8.4) has a result which reflects his actual strength very well. He isn't better than this.

Tochiazuma (11-4 / 9.1) exceed the expectations quite a lot.

Wakanosato (6-9 / 9.1) had the same expectancy as Azuma and did the opposite.

Hakuho (11-4 / 8.9) has exceeded his expected result according to the rating formula in every basho so far. The reason is simple: it takes some time for the rating to catch up when a new "player" is introduced. We have yet to see the limit for this boy.

Iwakiyama (8-7 / 7.3) did well to get his 8. It will be another struggle for KK next time.

Everyone else in the maegashira-joi (M1-M5) had an expectancy of between 6 and 7 wins for their bouts in this basho. Kakizoe (8-7 / 6.0) was just below 6, so kachi-koshi was good for him. Kotooshu (9-6 / 6.5) is new so his rating must be taken with a grain of salt. Kokkai (7-8 / 6.8) has been in makuuchi for a year and seems to have found his level of competition.

Most of the time the rikishi in this group will trade places with each other and get the occasional kachi-koshi good enough to get them to sanyaku. No wonder that the maegshira-joi is so hard to predict.

Kyokutenho (10-5 / 7.3) was rather low on the banzuke so 10-5 would perhaps not be so strange, if not for his hard schedule. According to ratings, he wasn't expected to do better than 7-8! His last two basho have affected his rating for the worse though. Tenho goes back to where he belongs, which is in the vicinity of Iwakiyama.

For makuuchi in general, the actual outcome were in many cases very close to the expected outcome. Some disappointments were Kotonowaka (4-11 / 6.1), Kotoryu (4-11 / 6.8) and Tokitsuumi (5-10 / 7.0), whereas there were few who exceeded their expected outcome with more than one win. No wonder this basho felt a little dull. Kasugao had 9-6 / 7.1. Ama had 8-5 / 5.7 in the 13 bouts he fought but his rating hasn't quite caught up yet.

In juryo Otsukasa (12-3 / 8.1) was way better than anyone expected, and Gojoro (4-11 / 8.0) was way worse.

#30 Asashosakari

Asashosakari

    Sumo is a very good thing. Though better with affordable video.

  • Active Members
  • PipPipPipPipPipPipPipPipPipPipPip
  • 16,389 posts

Posted 27 January 2005 - 05:58

Hakuho (11-4 / 8.9) has exceeded his expected result according to the rating formula in every basho so far. The reason is simple: it takes some time for the rating to catch up when a new "player" is introduced. We have yet to see the limit for this boy.

<{POST_SNAPBACK}>

The other side of this particular coin is very interesting as well, though. It's really quite amazing that Hakuho, being as new to Makuuchi as he is and with his ratings still in catch-up mode, was already expected to have 9 wins against full-on sanyaku opposition this basho.

#31 Doitsuyama

Doitsuyama

    Yokozuna

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 7,512 posts

Posted 27 January 2005 - 08:36

Hakuho (11-4 / 8.9) has exceeded his expected result according to the rating formula in every basho so far. The reason is simple: it takes some time for the rating to catch up when a new "player" is introduced. We have yet to see the limit for this boy.

<{POST_SNAPBACK}>

The other side of this particular coin is very interesting as well, though. It's really quite amazing that Hakuho, being as new to Makuuchi as he is and with his ratings still in catch-up mode, was already expected to have 9 wins against full-on sanyaku opposition this basho.

<{POST_SNAPBACK}>

Indeed. He already was #6 by ratings before the basho and is #4 with pretty much room to spare now. His rating is already on Ozeki level, and should even be underrated because of the fast ascent; the starting rating is 1800 and it takes a while to rise to 2346, Hakuho's 105 rated bouts aren't that much. Perennial Sanyaku Wakanosato had a slightly higher rating only once, before the Hatsu basho; now Wakanosato's rating of course is back to below-Ozeki level.

#32 Koukai

Koukai

    Jonidan

  • Regular Members
  • PipPip
  • 46 posts

Posted 22 April 2008 - 17:05

Doitsuyama, did you continue updating the ratings after 2005?

#33 Doitsuyama

Doitsuyama

    Yokozuna

  • Active Members
  • PipPipPipPipPipPipPipPipPipPip
  • 7,512 posts

Posted 22 April 2008 - 20:43

Doitsuyama, did you continue updating the ratings after 2005?

There are two follow-ups, the latest from 2006. They probably are a bit better than this initial thread, with more data, but the same principles.

I didn't come around to make a 2007 version due to some work with Sumo Reference, but it's likely that I will return to the strength analysis topic sometime this year.


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users