HenryK

Burn him alive.

This discrete drop in the Makushita division in 1967 from almost 200 to 120 must have been a major thing.

Interesting. Hence total Ozumo participation peaked in 1994, and has since than fallen back to the levels of the early 1980s. All the variation is in the lower division below Sd (unsurprisingly).

Below are what would be the current title holders. This is related but not identical to the ATP/WTA rankings, and in some cases I believe it reflects better a player's capacity to win a grand slam tournament (e.g., the Williams sisters are both Yokozunas but barely in the top 10 with the WTA). MEN Grand Champion (Yokozuna): Federer Champions (Ozekis): Djokovic (just promoted) Nadal (Nadal is currently Kadoban Ozeki and needs to reach the quarterfinal at the Australian Open to defend the rank) Roddick Junior Champion I (Sekiwake): Davydenko Junior Champions II (Komusubis): Ferrer Moya WOMEN Grand Champions (Yokozunas): Henin S. Williams V. Williams Hingis just retired, Davenport is half-retired, Seles may come back, Capriati is pondering a come-back... Champion (Ozeki): Ivanovic (Kadoban Ozeki) Junior Champions I (Sekiwakes): Kuznetsova (former Ozeki, back to Sekiwake with US Open final) Sharapova (just lost Ozeki status that she had held for more than 3 years) Junior Champions II (Komusubis): Bartoli Chakvetadze Jankovic Vaidisova

I for my part enjoy reading the sumo information on Wikipedia, and consider most articles well done. This said, a separate sumo wiki could be very attractive. And it seems to me there could and should be synergies between both projects: Sumo Wiki could take the existing Wikipedia articles as starting point, and Wikipedia could adopt future Sumo Wiki articles to the extent it considers this appropriate. And I'd be happy to help out (this said, there are many people here with a far more intimate knowledge of sumo).

Boy this looks nasty.

The Ozumo tournaments alternate between Tokyo and other locations. I believe there is a difference. Ask Takatoriki about Kyushu bashos. Probably not as great a variance as the effect of different surfaces in tennis, but still some. I don't understand the point being made here.... The point is that it's easier to obtain back-to-back victories if there are six tournaments (ozumo) a year as opposed to four (tennis). Kaio isn't a Yokozuna despite his five Yusho either. He is a very good Ozeki instead. See above -- 5 grand slams (Edberg won in fact 6, but never 2 in succession)) is probably worth a little more in relative terms than 5 yushos, as there are fewer GS tournaments. Newcombe is a special case, btw -- he had back-to-back victories in te pre-open era (Wimbledon and U.S. Open 1967) which would made him "Yokozuna", but then in the open era he was a bit erratic. The best "Ozeki" in my listing is Guillermo Vilas, btw -- 4 grand slams, but the best series he ever produced was semifinal-final-win in 1976/77. Also, Rafael Nadal would have made it to "Yokozuna" had he won one of his two Winbledon finals against Federer. He was close this year. Only permitting win-final-win as equivalent to win-win implies that Sanchez-Vicario and Davenport lose Grand Champion status. The rest remains unaffected. Thanks for your feedback. (Neener, neener...)

And here are the career high "Komusubis". TENNIS JUNIOR CHAMPIONS II (=KOMUSUBI) MEN 1 P. Gonzales 2 Ralston 3 Buchholz 4 Ruffels 5 Stolle 6 Goven 7 Emerson 8 Lutz 9 Froehling 10 Stone 11 Meiler 12 S.Mayer 13 Amitraj 14 Alexander 15 Case 16 Jauffret 17 Stockton 18 Dibbs 19 Giltinan 20 Pfister 21 Amaya 22 Dibley 23 Fibak 24 McNamara 25 Clerc 26 Frawley 27 Mayotte 28 Higueras 29 G.Mayer 30 McNamee 31 Roger-Vasselin 32 Arias 33 Scanlon 34 Teltscher 35 Smid 36 Testerma 37 Guenthardt 38 Nystrom 39 Masur 40 Chesnokov 41 Svensson 42 Cahill 43 E. Sanchez 44 Gunnarsson 45 Berger 46 Krickstein 47 P.McEnroe 48 Forget 49 Wheaton 50 Ferreira 51 Volkov 52 Larsson 53 Novakec 54 Eltingh 55 Woodforde 56 Rosset 57 Stoltenberg 58 Dewulf 59 Woodbridge 60 Escude 61 Kiefer 62 Kucera 63 Mantilla 64 Lapentti 65 Hrbaty 66 Melengeni 67 Squillari 68 Volchkov 69 Novak 70 Malisse 71 El Anayoui 72 Ancic 73 Ginepri 74 Youzny 75 Robredo 76 Gasquet TENNIS JUNIOR CHAMPIONS II WOMEN 1 Du Plooy 2 Shaw 3 Gunter 4 Schaar 5 Harris 6 Sawamatsu 7 Giscafre 8 Chmyreva 9 Newberry 10 Ekblom 11 Marsikova 12 Simon 13 May 14 Evers 15 Matison 16 Sawyer 17 Madruga 18 Stevens 19 Potter 20 Bunge 21 Smith 22 Rush 23 Yermaak 24 Bassett 25 Benjamin 26 Rinaldi 27 Lindqvist 28 Mc Neil 29 Provis 30 Cordwell 31 Porvik 32 K. Maleeva 33 Werdel 34 Schultz 35 Rubin 36 Mc Grath 37 Wiesner 38 Kournikova 39 Spirlea 40 Testud 41 Schnyder 42 Lucic 43 Stevenson 44 Dokic 45 C. Fernandez 46 Hantuchova 47 Zuluaga 48 Dechy 49 Likhotseva 50 Safina 51 Chakvetadze

If I start this why not doing it complete. So here are the junior masters. First level I (=sekiwake). There are many more men than women at this level, implying that the quality differences are relatively larger between female players (hence, a relatively larger share of titles etc. is snatched by the "Yokozuna" and "Ozeki" ladies) TENNIS JUNIOR CHAMPIONS I (=SEKIWAKE) MEN 1 Graebner 2 Okker 3 Gimeno 4 Crealy 5 Franulovic 6 Taylor 7 Richey 8 Anderson 9 Metreveli 10 Proisy 11 Orantes 12 Parun 13 Gorman 14 Pilic 15 Dent 16 Edmondson 17 Panatta 18 Solomon 19 Ramirez 20 Gottfried 21 Lloyd 22 Barazzutti 23 Marks 24 Pecci 25 Dupre 26 Sadri 27 Teacher 28 Warwick 29 Denton 30 Noah 31 Lewis 32 Jarryd 33 Pernfors 34 Leconte 35 Zivojinovic 36 Gomez 37 Ivanisevic 38 Pioline 39 Martin 40 Berasetegui 41 Muster 42 Washington 43 Rusedski 44 Bjorkman 45 Rios 46 Corretja 47 Philippoussis 48 Enqvist 49 Medvedev 50 Clement 51 Grosjean 52 Johansson 53 Costa 54 Nabaldian 55 Schalken 56 Schuettler 57 Verkerk 58 Coria 59 Gaudio 60 Henman 61 Davydenko 62 Puerta 63 Bagdhatis 64 Ljubicic 65 F. Gonzales 66 Haas 67 Tsonga 68 Ferrer TENNIS JUNIOR CHAMPIONS I WOMEN 1 Tegart 2 Bueno 3 Bowrey 4 Krantzcke 5 Masthoff 6 Durr 7 Hunt 8 Gourlay 9 Heldman 10 Mihai 11 Turnbull 12 Shriver 13 Nagelsen 14 O'Neil 15 B. Jordan 16 Walsh 17 Hanika 18 Durie 19 K. Jordan 20 Garrison 21 Kohde 22 Zvereva 23 Date 24 M. Maleeva 25 G. Fernandez 26 Huber 27 Coetzer 28 Tauziat 29 Suarez 30 Petrova 31 Jankovic 32 Vaidisova 33 Bartoli

Now here is the result. To me it looks eminently sensible, and I believe it would add another interesting twist to tennis if there existed such sumo inspired "career titles". In chronological order. TENNIS GRAND CHAMPIONS (="YOKOZUNA") MEN 1 Laver 2 Rosewall 3 Connors 4 Newcombe 5 Borg 6 McEnroe 7 Lendl 8 Wilander 9 Becker 10 Edberg 11 Courier 12 Sampras 13 Agassi 14 Federer TENNIS GRAND CHAMPIONS WOMEN 1 King 2 Court 3 Goolagong 4 Evert 5 Mandlikova 6 Navratilova 7 Graf 8 Seles 9 Sanchez-Vicario 10 Hingis 11 Davenport 12 V. Williams 13 Capriati 14 S. Williams 15 Henin TENNIS CHAMPIONS (="OZEKI") MEN 1 Ashe 2 Roche 3 Kodes 4 Smith 5 Nastase 6 Vilas 7 Tanner 8 Gerulaitis 9 Kriek 10 Curren 11 Mecir 12 Cash 13 Stich 14 Bruguera 15 Chang 16 Kafelnikov 17 Krajicek 18 Rafter 19 Korda 20 Moya 21 Kuerten 22 Norman 23 Safin 24 Hewitt 25 Ferrero 26 Roddick 27 Nadal 28 Djokovic TENNIS CHAMPIONS WOMEN 1 Jones 2 Richey 3 Casals 4 Wade 5 Melville 6 Morozova 7 Tomanova 8 Barker 9 Fromholtz 10 Jausovec 11 Stove 12 Ruzici 13 Austin 14 Jaeger 15 Sukova 16 Sabatini 17 M. Fernandez 18 C. Martinez 19 Pierce 20 Majoli 21 Novotna 22 Clijsters 23 Myskina 24 Sharapova 25 Dementieva 26 Kuznetsova 27 Mauresmo 28 Ivanovic I could add the junior champions if there is interest, but there are more than 200 of them. :-D

As a great admirer of the Ozumo ranking system, I've been thinking whether and how it could be applied to other sports. The closest cousin of sumo in terms of organization that I can think of is tennis. It's also one against one, and there are 4 major tournaments each year (not that different from ozumo's 6). Now I played a bit around with various ranking criteria since the beginning the open ear (1969). Here is what I came up with. Say a GS victory is worth 4 points, a final 3, a semfinal 2, and a quarterfinal 1. Then the following scores are needed to qualify for sumo-type ranks: TITLE Over 2 tournaments Over 3 tournaments GRAND CHAMPION 8 points 10 points CHAMPION 5 points 6 points JUNIOR CHAMPION 1 3 points 4 points JUNIOR CHAMPION 2 2 points 2 points In addition, a player can become a "career grand champion" if he or she wins 5 grand slam tournaments in his career, and a "career champion" if he or she wins 2 GS tourneys in his career -- even if he never passes the above thresholds. Justification below. In pratice, these criteria translate as follows -- a player qualfies for rank [X] if GRAND CHAMPION (=YOKOZUNA) a) he (or she) wins 2 subsequent grand slam tournaments b) he wins 2 tournaments interrupted by a final or seminfinal participation c) of 3 subsequent tournaments, he or she wins 1 and reaches the final in the 2 others d) he (or she) wins 5 grand slam tournaments during his career b) and c) are "equivalent achievements" to a). d) is different from sumo, but tennis is played on different surfaces, which makes it more difficult to repeat tournament achievements from one tourney to the other (and there af fewer tournaments). At a practical level, without d) Stefan Edberg and John Newcombe would not be grand champions. CHAMPION (=OZEKI) a) he or she wins one tournament and reaches the quarterfinal/semifinal/final in the next, and vice versa b) he or she reaches 3 subsequent semifinals; c) he or she reaches subsequently a final, a quarterfinal and a semifnal (in whichever order) d) he or she wins two grand slam tournaments during his or her career JUNIOR CHAMPION GRADE I (=SEKIWAKE) a) he or she wins a GS tournament or reaches the final b) he or she reaches subsequently a semifinal and a quarterfinal; or a final and a quarterfinal; or 2 semi-finals -- in any order c) he or she reaches a semifinal and 2 quarterfinals in three subsequent tournaments (in any order); or 2 semifinals interrupted by a quarterfinal JUNIOR CHAMPION GRADE II (=KOMUSUBI) a) he or she reaches a seminfinal b) he or she reaches two quarterfinals over the course of 3 tournaments

I beg to disagree. There is more difference between the best and the weakest maegashira, i think. Some rikishi were in maegashira for 50 basho or more and are now respected oyakata while other rikishi have a 2-13 makuuchi record. Fair point. I had more the sanyaku ranks in mind though.

I don't get you. All sekiwakes are 'preliminary' as it's a dynamic rank requiring constant KK. I don't understand what you mean by a 'full' sekiwake as opposed to a 'preliminary' one. Are you suggesting some sort of rank protection a la the ozeki rank for sekiwakes who have shown form ? If so, then I respectfully disagree :-D No. I am just pondering about a means to distinguish between rikishis that manage(d) to establish themselves at a junior sanyaku rank for extended periods on the one hand, and the one-off types on the other (the latter being responsible for the fact that almost every other maegeshira obtains sanyaku rank during his career). As the stats show, the sekiwake/komusubi distinction doesn't reallly provide this. While highest-rank komusubis are almost all one-offs, highest rank sekiwakes come from both camps. There is no other rank that groups together rikishis with that different achievements. To take an example, Kotonishiki and Hayateumi are both former sekiwakes, but their career achievements could hardly be more different. Kotonishiki's career Hayateumi's career

It's all the "one and done" career-high Komusubi skewing the numbers. :-D Of the 24 in the sample, 15 had just one basho at the rank. (The others: 4x two basho, 4x three basho, 1x four basho.) For comparison, the number of sanyaku basho by those 20 who topped out at sekiwake: 1 2 3 (twice) 4 (three times) 7 9 10 11 13 16 17 18 20 26 (twice) 27 34 Implying: obtaining a Sanyaku rank is (relatively) easy, keeping it is hard. .... and also suggests that the more important dividing line isn't between Sekiwake and Komusubi, but between established Sanyakus (overwhelmingly Sekiwakes) and temporary Sanyakus (mostly Komusubi, but also several Sekiwakes). Sekiwakes are a rather inhomogenous group. Maybe they should appoint Sekiwakes on a preliminiary basis first. Only when a rikishi has obtained the qualification for the rank a third or fourth time would he be converted into a "full" Sekiwake.

Sumo needs Asashoryu :-D ;-)

I was surprised reading this too, but according to my back-of-the-envelope computation your figure is entirely plausible. Take as benchmark the total number of rikishi as its designed to be -- about 790 (100+300+200+120+28+42). Then the average rank at any given time is 395, right at very top of the Jonidan division. I guess this the stat Asashosakari is after. But you are after the average rank per career and this must be lower, as people with low average ranks tend to have shorter careers. Thus underachievers they will be undersampled if you look at all rikishi at any point in time. As your computation shows, the difference between simple average and career-lenght weighted average is about 115 ranks. I understand and agree with your theory, but I believe it is inaccurate to assume that weighted average would be so close to the mean rank. If we assume that top-ranked rikishi have longer careers on average (not sure if it's true), then I I would expect "Asashosakari's weighted value" to be somewhat higher than the 395 you calculated (meaning higher on the banzuke). Hmmm.... The simple rank average at any given time gives every career the same weight, independent over whether it lastet 1 banzuke or 100. While for the average rank per career, one needs to weight ranks by the inverse of relative frequency of the rank holders's average career lenght. Implicitely, this is what Randomitsuki did. Thus I believe the simple average is in principle what Asashosakari is after. The effect of what you note -- top-ranked rikishis have longer career length -- is exactly the expressed by the difference between my statsitic -- 395 -- and Randomitsuki's -- about 510. A possible caveat are things like "people who aborted their career early could have had higher average ranks than they actually had", etc.. This is the question for the potential average rank. Computing this requires a full statistical model with rather strong assumptions (including: what rikishis cut their careers short and what didn't? For example, if Kaio retires rather than getting demoted to Sekiwake, did he cut his career short?) -- and I actually doubt one can get really reliable results.

I was surprised reading this too, but according to my back-of-the-envelope computation your figure is entirely plausible. Take as benchmark the total number of rikishi as its designed to be -- about 790 (100+300+200+120+28+42). Then the average rank at any given time is 395, right at very top of the Jonidan division. I guess this the stat Asashosakari is after. But you are after the average rank per career and this must be lower, as people with low average ranks tend to have shorter careers. Thus underachievers they will be undersampled if you look at all rikishi at any point in time. As your computation shows, the difference between simple average and career-lenght weighted average is about 115 ranks.

Don't forget that we have four active ozeki - we need only (the expected) two of them becoming yokozuna to get over 50%. Then again... maybe statistical expectation isn't everything. :-S I don't want to split hair, but isn't the correct calculation that since, say, 1999 10 rikishi obtained Ozeki status -- Chiyotaikai, Dejima, Musoyama, Miyabiyama, Kaio, Tochiazuma, Asashoryu, Kotooshu, Hakuho, Kotomitsuki -- but only 2 made it to Yokozuna, instead of 4-5 one would statistically expect?

It's all the "one and done" career-high Komusubi skewing the numbers. :-S Of the 24 in the sample, 15 had just one basho at the rank. (The others: 4x two basho, 4x three basho, 1x four basho.) For comparison, the number of sanyaku basho by those 20 who topped out at sekiwake: 1 2 3 (twice) 4 (three times) 7 9 10 11 13 16 17 18 20 26 (twice) 27 34 Implying: obtaining a Sanyaku rank is (relatively) easy, keeping it is hard.

Sanyaku as such appears to be somewhat less exclusive than one would expect. Ozeki is the really huge hurdle (and, at a lower level, Juryo).

It's still significant though...while only 46% of all debutants reach Sandanme, it's 55% for those who stuck around for more than a year. Bump it up just a little bit more to a year and a half, and it's already 64%. Absolultey. And I'm sure there are many ways how these stats could be refined.

Keep in mind that those who never advance beyond Jonokuchi tend to be rikishi only for the fleetest of moments...for those 512 rikishi mentioned above, their average career length was only 4.4 tournaments, with more than three quarters (388 of 512) sticking around for 6 basho or less. Sure. They are a relatively small part of the total (one-sixth), however -- 5 out of 6 Jonokuchis make it to Jonidan, but less than half make it to Sandame. The transition matrix above suggests that the relatively most difficult career steps are (i) from Sekiwake/Komosubi to Ozeki (28 percent), and (ii) from Makushita to Juryo (31 percent). All other career steps have a conditional probability of about 50 percent, with the exception of Jonokuchi to Sonidan (5 out of 6).

Inspired the above stats I have computed a little transition matrix. "8.2" in the field "Mk-S/K", for example, means that 8.2 percent of all Makushita debutants ended up obtaining Sanyaku rank in their careers. Ozumo Transition Matrix (in percent) [/td] To Y O S/K M J Mk Sd Jd From Jk 0.6 2.0 4.3 7.6 24.0 46.3 83.4 Jd 0.3 0.7 2.4 5.2 9.1 28.8 55.5 Sd 0.6 1.2 4.3 9.4 16.4 51.9 Mk 1.1 2.3 8.2 18.1 31.5 J 3.4 7.3 26.1 57.3 M 6.0 12.7 45.5 S/K 13.1 27.9 O 47.1

I have complete banzuke only from Kyushu 1974 on, so I arbitrarily decided to go with the 20 year mae-zumo frame from Aki 1974 until Nagoya 1994. I guess there won't be that many rikishi reaching new high divisions in the future out of this group despite 88 rikishi still being active. Now we get 3053 rikishi who entered the banzuke on jonokuchi and 36 more who started at makushita tsukedashi, not a bad sample size. The high divisions of the 3053 jonokuchi starters break down as follows: Jonokuchi 512 16.77% Jonidan 1147 37.57% Sandanme 688 22.54% Makushita 500 16.38% Juryo 94 3.08% Maegashira 62 2.03% Komusubi 19 0.62% Sekiwake 16 0.52% Ozeki 7 0.23% Yokozuna 8 0.26% Some sums: sekitori 6.75%, makuuchi 3.67%, sanyaku 1.64%. The makushita tsukedashi entrants in that time break down as 8 makushita, 6 juryo, 11 maegashira, 5 komusubi, 4 sekiwake and 2 ozeki (Asashio and Musoyama), 78% becoming sekitori. The only makushita tsukedashi ever to get to yokozuna was Wajima who entered a bit earlier in Hatsu 1970. Awesome. Thanks very much. This provided answers to several questions of mine. 1) The "average" Jonokuchi entrant will never make it to Sandanme (!) 2) Making it to Makushita is a huge achievement: less than a quarter of all rikishis get there. 3) Only a tiny elite ever gets to Sekitori: less than 8 percent (including Makushita entrants) 4) However, once you are there advancing gets relatively easier: about 60 percent of all Juryo debutants will make it to Makuuchi 5) And almost half -- 45 percent -- of all Makuuchi debutants will obtain a Sanyaku rank (!) 6) About a quarter of all Sanyaku debutants will go on to Ozeki 7) Finally, almost every other Ozeki (8 out of 17) will go on to Yokozuna (!)

That's not easy to answer - which mae-zumo time frame would you like exactly? "Past 20 years" is impossible to answer as many, many rikishi out of these will reach higher grounds in the future, but we can't know exactly how many. Dewanosato can tell you that you must really go a long time back for a definite answer, but then again this answer may be outdated... ok, which time frame? Would 1950-1990 make sense (taking your point that "past 20 years" results in a censoered sample)?