Obligatory health and safety warning: This is going to be perhaps some of the longest form writing committed to a forum page. Read at your own time and peril.
Natsu 2022 saw a really insipid performance from the ōzeki trio of Takakeishō, Shōdai, and Mitakeumi. Takakeishō very narrowly avoided kadoban, while Shōdai and Mitakeumi didn't; the former for the fourth time in his short ōzeki career. A bad performance once in a while might be excusable, but when two ōzeki going kadoban is already relatively uncommon (52 occurrences since 1949, when honbasho finally stayed at 15 days), the fact that two ōzeki have gone kadoban in 10 basho since 2017 is some cause for concern. In other words, 20% of the instances of two or more ōzeki going kadoban within the last 70-odd years happened within the last 5 years, which seems disproportionate, to say the least.
Handwringing about bad ōzeki performances is nothing new either - Gōeidō and Kotoshōgiku were (in)famously called the "kadoban twins" on Tachiai - but it's rarely been combined with detailed background analysis of the topic. One notable exception was in Haru 2021, when it looked like Takakeishō and Shōdai were at risk of kadoban on day 10; @Amamaniac and @Godango put in some solid work there to provide some stats and numbers on ōzeki performance or lack thereof, with other forum members chipping in. After Natsu 2022, @Yamanashi made the further observation about Kisenosato as either "the perfect ōzeki", or a potential yokozuna who could never "pull the string", which saw the following analysis from me and a request from @Gurowake to extend it further to an ōzeki's yūshō honour strike rate during his tenure. Serendipitously, @nagorastarted a thread a few days after Natsu 2022 on a rating system of their own design, which adds a further methodological arrow to the quiver in terms of analysing ōzeki performance.
After spelunking in a few rabbit holes, I'm proud to present what is probably the most comprehensive analysis of ōzeki performance to date. I should probably disclaim up front that I don't claim that this is the alpha and omega of such analysis, or that this is the final word on ōzeki performance. There are qualitative aspects of ōzeki performance that the following analysis does not capture, and there is at least one that I've had my eye on but decided to save for a future time. This is meant as a springboard to reflect on and possibly challenge various perspectives and narratives on ōzeki performance, and should be read in combination with the Forum's collective institutional memory on ōzeki as they were in their time, but it is hopefully the most thorough and least anecdotal springboard for a while yet.
This study encompasses all rikishi who have reached the rank of ōzeki since the rules for ōzeki demotion were changed in Nagoya 1969 to the current kadoban system, meaning that all yokozuna since Wajima (ōzeki promotion Kyūshū 1972) and all ōzeki since Kiyokuni (promoted Nagoya 1969) are included in this study. The kōshō system throws a bit of a wrench in this, but nothing that fundamentally invalidates the analysis. Also, definitionally, a zenkyu (0-0-15) will be considered the same as a make-koshi; only an actual kachi-koshi will be considered. There is room for debate about this, but I'll defend that by saying that an injured ōzeki is about as bad as an abunai ōzeki in the court of public opinion.
Naturally, SumoDB is the source of data for this project, although all transcription and scraping errors are mine - do feel free to point those out as and when you see them.
Due to its length, I'm posting this in 6 parts for safety's sake. If you see this while the rest of the posts are not up, bear with me for a bit. Also, I notice that some of the more voluminous tables are technically spilling over the borders of the post; I will reformat at a later date to deal with that but let me know if this is causing readability issues in the meantime (almost certainly on mobile, sorry!).
EDIT: And all posts are done - let fire!