Kyushu 2024 Promotion etc discussion

[Reonito 1,384]

12 hours ago, Asashosakari said:

Tochimaru (Ms20e 6-1)    Ms5  Otsuji (Ms14e 5-2)

So if they're generous with Onosho, Tochimaru is more likely to take the final spot in the promotion zone than Otsuji?

[Ack! 445]

On 30/11/2024 at 07:59, Ack! said:

I wonder if someone could explain the rank difference math Asashosakari used in his excellent analysis above? I struggle with finding a way to compare 15-bout results against 7-bout results, so this would be very helpful 

 

On 30/11/2024 at 09:02, Reonito said:

I believe he's just taking the difference between wins and losses for both 15-bout results and 7-bout results, using this difference as the expected number of ranks moved down/up, and essentially treating Ms1-Ms5 as though they were J15-J19 for this specific purpose.

I don't think this method would predict the relative ranking of J and Ms rikishi who end up in Makushita, or even the movements within upper Makushita, though I haven't done the analysis.

 

On 30/11/2024 at 10:10, Asashosakari said:

Yup. There are some interesting implications that feel very unintuitive, for example that Ms2w 4-3 would not be strong enough to force down J14e 7-8. I was rooting for Kotokuzan in his last match, but Miyagi finishing 4-3 would have been an acceptable outcome just to see that tested. (Ultimately it didn't matter because Daishoho collapsed even worse than Nabatame, and we wouldn't have got the Nabatame / Miyagi comparison anyway.)

My mistake: I meant to say the win/loss difference difference math.  The rank difference math is intuitive, but the win/loss difference math I cannot fathom.  Any explanation of that would be appreciated.

[Asashosakari 18,967]

3 hours ago, Reonito said:

So if they're generous with Onosho, Tochimaru is more likely to take the final spot in the promotion zone than Otsuji?

Classic 50/50 decision. 

[Reonito 1,384]

1 hour ago, Ack! said:

My mistake: I meant to say the win/loss difference difference math.  The rank difference math is intuitive, but the win/loss difference math I cannot fathom.  Any explanation of that would be appreciated.

Literally just subtract wins from losses. So 4-3 is +1, 5-2 is +3, etc. 5-10 is –5. As I said, this may not predict movement within makushita, but seems to predict the exchanges...

[Ack! 445]

19 hours ago, Reonito said:

Literally just subtract wins from losses. So 4-3 is +1, 5-2 is +3, etc. 5-10 is –5. As I said, this may not predict movement within makushita, but seems to predict the exchanges...

And now it's so obvious it smacks me in the face...  

[Asashosakari 18,967]

On 01/12/2024 at 04:51, Asashosakari said:

Banzuke guesses (...)

The additional stuff that's of even less interest:

Promotions from sandanme:
7-0   x1 Sd18w
6-1   x5 Sd5e, Sd10e, Sd25w, Sd30e, Sd42w
5-2   x8 Sd1e, Sd3w,  Sd9e,  Sd11e, Sd13w, Sd16w, Sd20e, Sd21w
4-3   x6 Sd2w, Sd3e,  Sd5w,  Sd7w,  Sd9w,  Sd10w
--------
     x20 (summary)

Shin-makushita for Hatsu:
Kotosato (Sd9w 4-3)
Anhibiki (Sd20e 5-2)
Yokomaru (Sd25w 6-1)
Kiyonohana (Sd42w 6-1)

Likely demotions from makushita:
intai x0
0-0-7 x2 Ms43e, Ms40e, (Ms20w)
0-7   x1 Ms44w
1-6   x2 Ms39e, Ms34e, (Ms29w)
2-5   x6 Ms59e, Ms57e,  Ms55w, Ms53w, Ms53e, Ms46w, (Ms38w)
3-4   x9 Ms59w, Ms58w,  Ms58e, Ms57w, Ms54w, Ms54e,  Ms52e, Ms50w, Ms50e, (Ms48e)
--------
     x20

At the risk of being as wrong as last time, I'll say again that there's really no other combination that makes sense. They did unexpectedly drop Ms40e 2-5 to save Ms50e 3-4 last time, but this time the distance between the last-out 3-4 and the next-out 2-5 is even larger, and the next-out 1-6 is also ranked very high. Perhaps the most likely curveball is a drop for the Ms20w 0-0-7, since he's set to go down to Ms60 anyway. Would be harsh, though.

Moving on...

Likely promotions from jonidan:
7-0   x2   Jd9w, Jd57w
6-1   x6-1 Jd1w, Jd22w, Jd40e, Jd44e, Jd52e, [Jd68w]
5-2  x12-4 Jd3w,  Jd5e,  Jd8w, Jd11e, Jd15e,  Jd16w, Jd19w, Jd21e, [Jd30w, Jd31w, Jd33e, Jd35w]
4-3  x11-6 Jd1e,  Jd2w,  Jd4w,  Jd8e,  Jd9e, [Jd10e, Jd13w, Jd14e,  Jd15w, Jd17e, Jd19e]
--------
     x31-11 (summary)

Shin-sandanme for Hatsu:
Goseizan (Jd5e 5-0-2)
[Takasu (Jd10e 4-3)]
Yonezawaryu (Jd22w 6-1)
[Yamane (Jd31w 5-2)]
Najima (Jd52e 6-1)
[Furuta (Jd68w 6-1)]

Likely demotions from sandanme:
intai x3    Sd24w, Sd64w, Sd78w
0-0-7 x5+1  Sd84e, Sd82w, Sd65w, Sd44w,  Sd39w, [Sd29w]
0-7   x0+1 [Sd38w]
1-6   x4    Sd83e, Sd74w, Sd62w, Sd53w, (Sd44e,  Sd37e)
2-5  x10+3  Sd90e, Sd86e, Sd84w, Sd82e,  Sd80e,  Sd76w, Sd75e, Sd68w, Sd66w,  Sd65e, [Sd61w, Sd60w, Sd56e], (Sd53e)
3-4   x9+4  Sd89w, Sd89e, Sd88w, Sd87w,  Sd85e,  Sd80w, Sd77w, Sd73w, Sd72w, [Sd71e,  Sd68e, Sd67w, Sd67e], (Sd63w)
--------
     x31+9

Here's where it gets tricky this time, thanks to the reduction of the sandanme division from 90 ranks to 80 that was recently confirmed. Superficially you might think that that's just going to mean that all the promotion rank cutoffs will be 10 higher, e.g. promotions will be given to all 4-3's down to Jd10 instead of the standard Jd20, and so on. They could do that, but it's not how past divisional size changes were implemented in practice. Instead they appear to create the new banzuke as they normally would, and then just re-number the ranks as needed.

The issue introduced by that approach: Only rikishi right at the floor of a given score's promotion zone are actually guaranteed an equivalent full-size promotion. Somebody who's 4-3 at Jd20 obviously needs to be promoted (at least) 20 ranks, because otherwise he wouldn't be going to sandanme at all. That's not true for a rikishi who is getting promoted from, say, Jd10, and consequently the last sandanme cut after Haru 2022 saw two rikishi stuck in jonidan despite rank/record combinations that were at least 10 ranks better than the standard floors (Jd10w 4-3 -> Jd2w, Jd58e 6-1 -> Jd1w).

My predicted rank movements do include one such loser of the special circumstances, Takasu (Jd10e 4-3), who would also miss out on his sandanme debut if I'm right. Two other would-be debutants, Yamane and Furuta, are definitely going to miss out because they're just too close to the normal promotion floors. The lists above are such that the ranks listed in [angled brackets] are those that are affected specifically by the sandanme size change, either not getting promoted from Jd to Sd, or additionally getting demoted from Sd to Jd. Altogether I have 11 rikishi not going up who normally would, and 9 rikishi going down who normally wouldn't, totalling the 20 eliminated spots. Chances are high that some part of this projection is wrong, though.

And lastly...

Jd/Jk split:
 186 Jd
  28 Jk
-  6 intai
-  2 >Bg
+ 11 Mz
+ 20 Sd adjustment
----
 237 = 118.5 ranks

Likely to be in Jk:
Mz       x11
Jk 0-7   x1  Jk14w
Jk 1-6   x2  Jk11e, Jk10e
Jk 2-6w  x1  Jk9w
Jk 2-5   x4  Jk9e,  Jk6w,   Jk6e,   Jk1w
Jk 3-4   x6  Jk10w, Jk7w,   Jk7e,   Jk4w,  Jk3e,   Jk2w
Jd 0-0-7 x5  Jd85w, Jd81w,  Jd58w,  Jd48e, Jd42e, (Jd27e)
Jd 0-7   x2  Jd67e, Jd53e
Jd 1-6   x2  Jd87w, Jd84e, (Jd70e)
Jd 2-5   x3  Jd93w, Jd90e,  Jd85e, (Jd81e, Jd80e)
Jd 3-4   x0 (Jd91w, Jd91e,  Jd89w,  Jd88e)
-----------
        x37 => 100 Jd, 18.5 Jk

There's no particular "banzuke logic" reason to expect exactly this split, but it would feel appropriate to me if they tried to establish a 100+ rank jonidan after the sandanme cut again, and there's arguably not much of a point in going beyond exactly 100 since the jonokuchi remainder would be very short as is.

Small bonus to close things out, a jonokuchi banzuke prediction (I doubt I'm going to make that a regular feature):

Fujinonami (Jd85e 2-5)    Jk1   Kobayashi (Jd42e 0-0-7)
Chiyooume (Jd53e 0-7)     Jk2   Koga (Jd90e 2-5)
Azumayama (Jd93w 2-5)     Jk3   Amamidake (Jd84e 1-6)
Wakasei (Jd87w 1-6)       Jk4   Daikisho (Jd48e 0-0-7)
Nakayama (Jk2w 3-4)       Jk5   Wakaarata (Jk3e 3-4)
Soga (Jk4w 3-4)           Jk6   Kikuchi (Jk7e 3-3-1)
Tenkoyama (Jk7w 3-4)      Jk7   Itakozakura (Jk10w 3-4)
Chiyorozan (Jd67e 0-2-5)  Jk8   Yumenofuji (Jd58w 0-1-6)
Katsunishiki (Jk1w 2-5)   Jk9   Moriurara (Jk6e 2-5)
Kyonosato (Jk6w 2-5)      Jk10  Yamada (Jk9e 2-5)
Sadanojo (Jk9w 2-1-5)     Jk11  Hara (Jk10e 1-6)
Sawaisamu (Jk11e 1-6)     Jk12  Nishikimaru (Jd81w 0-0-7)
Amanofuji (Jd85w 0-0-7)   Jk13  Gonosho (Mz 3-0)
Goseimaru (Mz 3-1)        Jk14  Okada (Mz 3-1)
Wakaikazuchi (Mz 3-2)     Jk15  Ikazuchihayate (Mz 3-2)
Ideno (Mz 3-2)            Jk16  Tsuyasato (Mz 2-4)
Fujihara (Mz 2-3)         Jk17  Daiko (Mz 1-4)
Arise (Mz 1-1-4)          Jk18  Wakasasaki (Jk14w 0-5-2)
Higohikari (Mz 0-4-2)     Jk19  ---

Edited Monday at 07:10 by Asashosakari

[Gurowake 4,016]

10 hours ago, Asashosakari said:

Instead they appear to create the new banzuke as they normally would, and then just re-number the ranks as needed.

Given that we know that's not how they do banzuke normally, it's odd that they would do this when they're renumbering ranks, especially since they normally have set promotions from each division to the next.  I do grant that the one think that's consistent about their approaches in unusual situations is that it's inconsistent.

Edited Monday at 17:03 by Gurowake

[Asashosakari 18,967]

4 hours ago, Gurowake said:

Given that we know that's not how they do banzuke normally (...)

I'm not entirely sure what you meant here.

[Gurowake 4,016]

2 hours ago, Asashosakari said:

I'm not entirely sure what you meant here.

I meant that they don't just first create an ordered list and then divide it up into divisions, which is what they apparently do when the divisions change size.

[Asashosakari 18,967]

2 hours ago, Gurowake said:

I meant that they don't just first create an ordered list and then divide it up into divisions, which is what they apparently do when the divisions change size.

I think we've been talking past each other here. What they will have done is create a regular sandanme ranking to rank 90 that then just gets cut off at 80 after the fact. I'm not quite sure what difference in approach you're trying to postulate, but it's not there.