Evaluating Ozeki Careers

Kaio, the gold standard for Ozeki

Commenting on my post about Ozeki runs, SMB asks:

I assume that the people deciding on Ozeki promotions are aiming to promote rikishi who are likely to make long-term Ozeki and potential Yokozuna. You included some examples of the outcomes of Ozeki promotions, but it would be interesting to see more analysis of how successful the Ozeki promotion decisions have turned out to be. In other words, do the decision makers largely get it right, if they are aiming for long-term Ozeki?


This is a great question. Of course, what counts as a long-term or successful Ozeki is necessarily somewhat subjective, but we can look at the history to get an idea. As usual, I will limit the analysis to the six annual basho era from 1958 to the present. Sixty rikishi were promoted to Ozeki during this time. How have they fared?

Reaching The Top

A clear career goal for Ozeki is making it to Yokozuna, sumo’s highest rank. Of the 60 most recent Ozeki, 26 have made this final ascent. Five who have not are still active, although the chances that Kotoshogiku and Terunofuji regain the rank of Ozeki, much less become Yokozuna, seem exceedingly remote, and I would put the likely expected number of promotions from among Goeido, Takayasu, and Tochinoshin at one. In any case, the odds that an Ozeki takes the next step are close to 50:50.

While it could be argued that making Yokozuna always counts as success, there is one obvious exception: Kitao. He became the 60th Yokozuna, under the shikona Futahaguro, after only four basho as Ozeki and despite not winning a yusho. He failed to do so in his eight more tournaments at the top rank as well, becoming the only Yokozuna not to win a championship, and was forced to leave sumo in disgrace.

Many of the most successful Yokozuna have treated the second-highest rank as a brief waypoint on the climb to the top. The nine Ozeki not named Kitao who spent between 3 and 7 basho at the rank have 191 yusho among them, and seven of them are all-time great Dai-Yokozuna: Taiho, Wajima, Kitanoumi, Chiyonofuji, Akebono, Asashoryu and Hakuho. Of the three other modern Yokozuna with double-digit championships, Takanohana had to wait 11 tournaments (with 5 yusho!) for promotion, Kitanofuji took 21, and Musashimaru is tied for the most tournaments at Ozeki before Yokozuna promotion with 32, as well as the most Ozeki yusho with 5. The average number of basho at Ozeki before promotion is 13.5, or a little over two years.

Interestingly, in light of the oft-discussed “two consecutive yusho or equivalent” guideline for Yokozuna promotion, fully half of the 26 promotions happened with either zero (Kitao and Mienoumi) or one yusho as Ozeki, and only 11 followed consecutive championships. For more on the rather fluid and ever-changing criteria for Yokozuna status, I highly recommend this terrific video by Chris Gould:

You can skip to the part about modern promotion guidelines at 8m 15s, but the whole thing is well worth a watch.

The Terminal Ozeki

What of the 29 (retired) rikishi for whom sumo’s second-highest rank represented the pinnacle of their career? They range from Daiju, who lasted 5 undistinguished basho at the rank in 1973-74, to Kaio, who set the standard for Ozeki longevity and excellence with 65 basho (tied for the most with Chiyotaikai) and 4 yusho, the most among Ozeki who did not become Yokozuna. On average, an Ozeki career has spanned just over 27 basho, or 4.5 years. The time spent at the rank drops to around 21 basho, or 3.5 years, if we include those who vacate it through Yokozuna promotion. Given that there have been 60 Ozeki promotions in just over 60 years, 3.5 also represents the expected number of Ozeki on the banzuke at any one time, so we are a tad below average with three at the moment.

How many of the 29 can we regard as “successful”? Six lasted 12 basho or fewer at the rank, with zero yusho. These are pretty clearly misses, for one reason or another. On the other end of the distribution, there are eight “super-Ozeki” who held the rank for 36 or more basho and won at least one yusho. In addition to the aforementioned Kaio and Chiyotaikai, this group includes such notables as Takanohana I, the father of two future Yokozuna, Konishiki, the first foreign Ozeki, and Kotooshu, the first European to raise the Emperor’s Cup. The other 15 can be regarded as average Ozeki, serving between 2 and 6 years and often picking up a yusho along the way (Wakashimazu managed 2 yusho, and Tochiazuma won 3). So, if we regard these as “successful”, then the decision makers “get it right” almost 90% of the time; if we raise the bar to either lasting longer than average at the rank or being promoted to Yokozuna, this rate drops to a little over 70%.

Does The Quality Of The Ozeki Run Matter?

As noted in my previous post, 18 of the 60 Ozeki reached this rank after recording fewer than 33 victories in the 3 basho before promotion, 14 recorded exactly 33, and 28 recorded more. The average is just over 33 wins. We can ask if the number of wins during the Ozeki run correlates with subsequent success at the rank.

How do we measure success? We can try a couple of ways. First, does the number of wins predict which Ozeki go on to make Yokozuna? The answer is “no”—the correlation is actually slightly negative, with 33 wins for those who eventually make the next level, and 33.4 for those who do not. Ten of the 18 rikishi promoted with fewer than 33 wins reached the highest rank. Next, we can look at the length of the Ozeki career among those top out at the rank. Here, there is a weak positive correlation (0.18) with the number of wins, but it’s not statistically significant.

Interestingly, the one metric that does show a moderate correlation (0.39) with the number of wins during the Ozeki run is career yusho won by those who make Yokozuna. Only two of the 10 Dai-Yokozuna were promoted to Ozeki with fewer than 33 wins, while six recorded more than the target number. The future greats usually have no trouble racking up wins on their way to Ozeki promotion, just like they tend to pass through the rank quickly.

How Do The Active Ozeki Stack Up?

With the historical numbers in hand, I’ll close with a brief look at the active Ozeki (or former Ozeki). Assuming that Kotoshogiku’s Ozeki days are behind him, his 32 basho and 1 yusho would place him somewhat above average among those who never made Yokozuna. Were his career to end today, Goeido, with 27 basho at the rank and one yusho, would already grade out as pretty much exactly an average terminal Ozeki, and of course he will continue to accrue service time, so the scorn often heaped on him is undeserved. With his 14 basho as Ozeki, the sad tale that is Terunofuji falls just outside the bottom six. And we can all hope that the stories of the Ozeki careers of Tochinoshin and Takayasu are far from finished.

25 thoughts on “Evaluating Ozeki Careers

  1. Maybe it would be fairer to deduct the number of kadoban basho from the total number of basho if we are talking about career length as a proxy for career strength.

    Another correlation I would like to explore is the quality of wins during Ozeki run (wins over Ozeki and Yokozuna) with the “strength” of the Ozeki career, rather than number of wins.

    • I agree those would be interesting analyses, but I’m not sure how to get those stats without doing it manually.

      • At first I thought the kadoban query would be easy: basho 1 at Ozeki, less than 8 wins, basho 2 still at Ozeki. But then I realized that Chiyotaikai enjoyed a lot of kosho-seido. He may have single-handedly caused the abolition of that system. It looks like under today’s criteria he would have broken the record for number of times at Ozekiwake, as each time he followed the kyujo with a double digit (not to mention yusho) performance. :-)

        Kaio also had one of those, and so did Tochiazuma… though he did get to be Ozekiwake a couple of times once the system was abolished.

        Still, those occasions still count as kadoban.

        • I’m game to analyze any data that someone can pull from the db, or that we can crowdsource. As soon as it’s in a google doc or similar, I’m good to go ;)

          • I’ve been thinking about writing a post with similar origins but that goes in different directions so this post is a great jumping off and reference point, thank you!

            I would largely agree with Herouth’s comment though, and I think this is why qualitative analysis both in terms of data but also just from watching a hell of a lot of sumo (ie. the eye test) and articulating is important. The comments re: Goeido and Terunofuji really resonated for me because Goeido is largely thought of as disappointing because of his potential (though his poor decision making is more or less reflected in the data with the number of kadoban) – whereas I’ve always felt that Terunofuji was sort of romanticised while active on account of performance in a handful of basho, but that history would probably judge him fairly harshly on account of his actual body of work.

            • Thanks Josh, I completely agree that my limited quantitative analysis misses a lot, but (1) I’m a numbers guy and (2) I haven’t been watching sumo long enough to have much of a historical perspective for the “eye test.” Lots more to explore for sure; looking forward to your post!

              • I read an article I can’t seem to locate written about a decade ago about amateur sumo. In it the author reports on a conversation he had with two highly skilled amateur sumo coaches in 2006 about who might be the greatest yokozuna of all time. The author thought Asashoryu might be it but the sumo coaches were of the opinion that recently promoted ozeki Hakuho would soon prove the better wrestler. Experts can perceive what even avid spectators can’t.

              • All good – not at all a criticism by the way. Without wishing to toot our own horn too much, I think one of the cool things about what we are creating here is that when it comes time for the wave beyond the Takakeisho’s of the world and we’re looking several ozeki promotions down the line, we’ll all be able to reference posts like this and this kind of work with numbers to have context in the future and speak from the experience of having gone through it, which makes everyone’s analysis better.

  2. Very nerdy and extremely enjoyable post! And very cool that you linked to the Chris Gould-video. That guy has a lot of interesting content.

  3. I don’t think of Kotooshu as much of a “super-Ozeki”. He reached Ozeki very quickly (around 3 years from Mz) but never got anywhere near the level of his Ozeki promotion run again consistently, only posting a Jun-yusho once in addition to his Yusho. Super-Ozeki to me are more along the lines of those that put up multiple Ozeki-promotion quality runs (or close to it, or at least one really good run) while Ozeki. Kaio and Chiyotaikai did it multiple times in their prime. Kisenosato did it 4 non-overlapping times before being promoted, and many more if you’re allowed to overlap them. About half of Ozeki never manage to do it again, and those that do manage it multiple times are usually talked about as being the best Ozeki never promoted. Terunofuji and Takayasu were/have been close twice with not much time at Ozeki, while Kotoshogiku and Goeido were/have never (been) close.

    • That’s an interesting perspective, and I might even try to look at it quantitatively if I can figure out the right db queries.

    • The easy search was for overlapping runs of 33+ wins in 3 basho as Ozeki without Yokozuna promotion. Man, Musashimaru was a monster by this metric before he finally got promoted. Konishiki and Takanonami are the standouts among those who were never promoted. Looks like Kotoshogiku did it once (when he won his yusho, but with a 8-6-1 middle basho) while the other active guys have not. And Ashifuji and Takanohana put up runs of 40 in 3! http://sumodb.sumogames.de/Query.aspx?show_form=0&columns=5&n_basho=4&sum_wins=33&show_sum=on&group_by=rikishi&form1_rank=o&form1_year=1958-2018&form2_rank=o&form3_rank=o&form4_rank=O

      • Although, as much as I loved the guy (the first rikishi I truly considered myself a fan of) – Takanonami’s numbers were just massively inflated by the Futagoyama heya advantage of that era. I don’t remember all the details exactly now, but a while ago I was trying to “re-calculate” the likely career winning rates of the various top Futagoyama rikishi if they hadn’t been able to avoid wrestling each other, and IIRC Takanonami ended up getting discounted by more than a full win per basho for having gotten tons of mid-ranked maegashira opponents through his ozeki career instead of Takanohana/Wakanohana/Akinoshima/Takatoriki/etc..

        The same procedure ended up docking Takanohana about half a win per basho for his time as yokozuna, incidentally – almost exactly the real-world difference between him and Akebono (12.2 vs 11.7 wins per basho, disregarding absences).

    • I agree with that or even of not consecutive, but simply the number of basho double digit wins. I think one thing that really set Kisenosato apart from Goeido, and made the later look more a dissapointing Ozeki, ist that he consistently put up double digit wins. If you take the 34 basho from the start of his Ozeki run till his promotion to Yokozuna, he had 1x 7-8, 6×9-6, 10×10-5, 8×11-4, 3×12-3, 5×13-2 and 1×14-1.
      Goeido on the othe rhand has 30 basho including his Ozeki run and managed a mere 6 times double digit wins, two of those during his Ozeki run, so this leaves only 4 double digit bashos out of 27 Ozeki appearances.
      Imho a good Ozeki is one, who challenges the Yokozuna corps, even if he were never to be on an actual Zuna run, but often there to keep the basho open/interesting till final weekend.
      Kotoshogiku had 9 double digit basho as an Ozeki in 32 basho and 3 during his Ozeki run.
      Terunofuji is a complete outlier. Two of his3 tournaments during the Ozeki run he spent at Maegashira rank, the first one being a mere 8-7. During his whole career he had a strong run of 4 tournaments in 2015(48wins) and a short comeback of two basho with 13 resp. 12 wins in 2017. The rest of the time was injury ridden. The 4 times he went Kadoban from January 2016 till January 2017 he never did so with less than 11 losses.

      Does that mean Terunofuji was a bad choice? I don’t think so. Same as Kisenosato wasn’t a bad promotion for Yokozuna. In both cases it was just unlucky cirucmstances that their health deterioted dramatically shortly after promotion never letting them live up to their potential.

      Kaio btw, for all his longevity was a strong Ozeki only from 2000-2004. In the 29 tournaments starting with his Ozeki run (8, 14, 11w) he managed double digit wins in 21 tournaments, including 5 yusho. Of the 8 non double digit ones, 2 were makekoshi, one was a bare 7-8 makekoshi, 4 were makekoshi were he went kyujo for significant parts and one he sat out due to injury. in 2005 he exchanged makekoshi/kyujo tournaments with 10-5 basho. From January 2006 till July 2011 he managed exactly two double digit tournaments (10 and 12 wins) and 5 kadoban. 15 times during that period he finished 8-7 including all basho in 2009. Looks to me as after 2004 he was moe or less a non-factor, much like Goeido. I haven’t watched sumo during his career and can just guess that somewhere in 2004 he suffered a career limiting injury.
      While I think that Kaio is a “Dai-Ozeki”, I would base that on the first years of his Ozeki reign, not the longevity he showed, even though that is quite an accomplishement as well.
      However I think his career perfectly illustrates that the number of basho as an Ozeki shouldn’t really be the criteria. If you only look of the years from 2005 on that would make 39 tournaments, which would make him a super Ozeki by your definition, while I would rather say he was a sub par one during that period.

      • Great perspective and info; you should write a post ;) I think it is a key point that short/poor Ozeki careers don’t necessarily mean those promotions were mistakes, as the results could be due to injuries sustained after the promotion, and these cannot be predicted.

      • It was more that by 2005, Kaio had been wrestling at the highest levels of sumo, i.e. against the joi in every basho, for a full decade, and had the amount of accumulated injuries that one would usually expect to see from a 32-year-old who has gone through that. (See Kotoshogiku for a very similar case.)

        At some point he appeared to resign himself to the fact that exceptional longevity was the only thing he realistically had left to achieve, and he wasn’t shy about talking about how broken down his body was, including admitting that he needed daily massages from his physiotherapist/ex-pro wrestler wife during the basho just to be able to mount the dohyo day after day. His sumo also became extremely conservative and geared towards avoiding injury, leading to that plethora of minimum kachikoshi records. He probably could have won more bouts in many tournaments, but he wouldn’t have lasted as long. Having been arguably the top fan favourite of that era, the Japanese fans largely didn’t seem to mind.

        In the end, he managed to hang on long enough to break Chiyonofuji’s career wins record, and it was very much not a coincidence that he retired less than a week later.

        • Thanks for the great insight. I didn’t really pay attention to his age to be honest. As I said, I didn’t follow sumo during his career. In any case, he has been 38 at Fukuoka 2010, where he finished with 12-3. Don’t think anyone before or after him produced such a score at this age.


This site uses Akismet to reduce spam. Learn how your comment data is processed.