Re: Finding the peak

["Josh Reyer" <Reyer@benchsumo.zzn.com>]

Nice, uh, "sabermetric" analysis, G. Jay!  Interesting that sumo 
wrestlers peak at the same age as baseball players.

Do you have the data resources to do a comparison study of rikishi 
who were active from c.1960 to c.1980?  It would be interesting to 
see if advances in nutrition and training have provided a less steep 
decline phase for current rikishi compared to rikishi of past eras.

I would also suggest that this is a good argument for earlier 
promotion (and thus better prognostical promotional standards).  If 
you wait to promote a rikishi when he's 30, like as not he's not 
going to continue the same level of performance the got him the 
promotion.  (Kaio, anyone?)

Josh Reyer

---- Begin Original Message ----

From: "G. Jay Walker" <walker@nhrc.navy.mil>
Sent: Tue, 10 Dec 2002 15:36:17 -0800
To: sumo@sun01pt2-1523.statgen.ncsu.edu
Subject: Finding the peak


(warning - pseudo-stat geek post)

At what age do rikishis peak?  How rapidly do they decline after 
reaching their peak?  What percent fight effectively into their 30's?

I use a rank/record type of rating system occasionally.  It lacks a 
bit of the accuracy of the bout-by-bout systems others have devised, 
but it takes a lot less time to set-up and calculate.  As an 
analytical tool, it wouldn't qualify as a sharpshooter's rifle, but 
it does kind of resemble an elephant gun.   For more precise 
questions needing a sharpshooter's accuracy - say like comparing the 
careers of two rikishi - you may want to indeed grab your 
sharpshooter. But if you're trying to answer a big broad question - 
like at what age do rikishis peak? - than maybe it's okay to use an 
elephant gun to bring down an elephant.  

Using my rank/record system, I've developed lists of the top 40 
rikishi going back quite a few years.  For any given year, I awarded 
8 points to a rikishi if he finished in the top 5 for the year, 7 
points if he ranked 6th through 10th, down to 1 point for those 
ranked 36th to 40th.  Those ranked below 40th received 0 points.  So 
Kotonowaka, who ranked 18th this past year, earned 5 points at age 
34.  Daizen, ranked 34th, earned 2 points at age 37.

The study group was all the top-division rikishi born in the 1960's, 
starting with Asahifuji (born July 6, 1960) through Asanosho (born 
December 23, 1969) - a total of 73 rikishi.  They started reaching 
the top 40 in 1981 and a few still remain in the top 40 through this 
year (Kotonowaka, Akinoshima, Daizen, Asanowaka).  However, all have 
now reached at least 33 years, so they shouldn't present a problem in 
determining the peak.

The three columns below show the age, the total number of 
rikishi  who ranked in the top 40 at that age (out of the 73 in the 
study) and the total number of points they earned at that age:

Age
20	 6	 22
21	 9	 48
22	18	 84
23	28	123
24	33	153
25	44	191
26	44	222
27	48	225
28	42	189
29	37	159
30	25	107
31	19	 75
32	11	 48
33	10	 41
34	 8	 23
35	 6	 17
36	 4	 11
37	 2	  3
38	 0	  0

(Note - a few more points might be added to the 33-and-over totals if 
Kotonowaka, Akinoshima and Asanowaka can stay in the top division.)

Perhaps these results don't reflect any more that what an astute 
observer would pick up from a few years of watching sumo, but they do 
quantify such observations.  The peak is usually reached at 26 or 
27.  Most young rikishi don't make their mark until 23 or 24.  There 
is not a 'plateau' after reaching the peak, but in fact a significant 
decline between the ages of 27 and 30.  For every rikishi performing 
in the top division at age 27, only about half will be there at age 
30; only about 20% at age 33.  

If form holds (and there are exceptions), collectively one would 
expect the 76'ers (those born in 1976, including Chiyotaikai, 
Tochiazuma, Kotomitsuki and Wakanosato) not to show that much 
improvement in their remaining careers.  While there may be an 
individual breakthrough or two, at age 26 they are more likely about 
as good as they'll ever be.  On the other hand, for Asashoryu at age 
22^Å well, you can look at the table and draw your own conclusions.

Another study using a more accurate weapon than an elephant gun might 
alter a few of the results found here in some modest ways. And as 
time moves on, results from rikishis born in the 1970's might vary 
 from these results (there are some early indications that the 27-to-
30 decline is becoming less pronunced).  But I don't think they would 
change any of the main conclusions.

-George W.



---- End Original Message ----




If a first you don't know what to do... henka
___________________________________________________________
Get your own Web-based E-mail Service at http://www.zzn.com