3 Comments
User's avatar
glacier's avatar

Are you attempting to find the 50th percentile or the average? Your first model indicates the former, but the rest indicates the latter.

That of course impacts the estimate by quite a bit - my first pass without reading anything was a median/50th percentile of 10 and an average of 15.

My first reaction to an estimate of 26 is that I'd sell a lot of size there - it's not unreasonable I suppose but it is basically saying that Onosato will be a top 5 rikishi(and even Asashoryu didn't get there; so even this is assuming Onosato has very low/no probability of getting suspended which I guess I believe)..

Neither Takanohana or Akebono got there - and it wasn't even mainly because of overlapping primes; Akebono was basically done winning yusho when Takanohana was 21.

Of course including injuries will bring this number down into more reasonable territory but I also think you are just overrating Onosato - his last 7 basho are nice but certainly nothing historical: https://sumodb.sumogames.de/Query.aspx?show_form=0&columns=7&sum_wins=75&sum_range=7&show_sum=on&form1_rank=Y-M4&offset=550

He's good no doubt but have we really seen something that indicates he is as good as Hakuho/Taiho/Chiyonofuji?

Expand full comment
Ozeki Analytics's avatar

Yeah that's absolutely fair! That said the goal here was less to super accurately model the outcome and more to start from first principles and see how we might go from just tossing out a random number and towards having a more structured process that lay people and obsessives alike could work with or build upon for the latter.

And I'll definitely cop to being optimistic on him, perhaps overly so, but more than happy to be upfront about that. I'm sure in a year or two when the new Yokozuna glow has shined off a bit it'll be fun to revisit this and see how the predictions look. Cheers!

Expand full comment
glacier's avatar

This was a fun exercise for me looking up how his last 7 basho stack up because I hadn't realized how close it was to KTZ's best run(77 in 7 to Onosato's 79). Of course KTZ has been at the top for a while whereas Onosato is working off a sample size of exactly 1; so the gap is larger than that stat shows.

It's definitely possible that his bad bashos were because of yusho/promotion hangovers or whatever and he's a consistent 12/13+ win guy. On the other hand, you could also say that he's turning 25, he's miles behind Hakuho/Asashoryu/Taiho at that age, and any improvement will be cancelled out by the rest of the division adapting to him.

Expand full comment