Asashosakari

Banzuke for Natsu 2022

Recommended Posts

Posted (edited)
13 minutes ago, Yarimotsu said:

As maglor referenced, this trend isn't actually strange when you look at how far down the division extended at that particular point in time. My working theory for why we saw midorifuji and kagayaki instead of chiyonokuni and chiyomaru actually involves the separation of ishiura and meisei on the banzuke - I think that for whatever reason, the committee decided to punish rikishi for not sticking around to fight all 15 days (even when chronically injured).

Not sure that explains the distinction between the two Chiyos, does it? Both "scored" the same wins (albeit Chiyonokuni had withdrawals rather than losses for some of them) and both were dropped from the division from the same rank. In fact, the prevailing wisdom before the banzuke dropped would have squared with your theory, since most people would have kept Chiyomaru, but not now.

You might be on to something re Meisei and Ishiura, but if there's one thing the banzuke makers are consistent about, it's being annoyingly inconsistent with these over a long period of time, and sometimes even in the same banzuke - my personal grievance is with the separation of the 5-10 mid-maegashira trio when there was no apparent reason to do so. My working theory for that is that they put the KKs where they think they should go and the MKs fill the gaps, but no doubt someone will come and put holes in this too (for one I think there's too many MKs in the lower half of the banzuke to not need to bump some KKs up or down to make it work).

Edited by Seiyashi

Share this post


Link to post
Share on other sites
On 24/04/2022 at 17:21, Asashosakari said:

Ishiura getting demoted half a rank more than Meisei with a record one win better. :-S No way that's just a joi bonus for Meisei

And yet somehow Ura ended up higher ranked than Wakamotoharu, which defies pretty much any logic other than a full win bonus for full joi schedule.  I had Ura ahead of Takarafuji since the joi bonus is very often more than a half rank, but putting Ura ahead of Wakamotoharu just seems absolutely insane to me.

  • Like 1

Share this post


Link to post
Share on other sites
Posted (edited)
56 minutes ago, Seiyashi said:

my personal grievance is with the separation of the 5-10 mid-maegashira trio when there was no apparent reason to do so.

There was a very good reason to separate them - just not the one they went with.  However, the fact that you thought they should all be next to each other while I thought Okinoumi should be split from Chiyoshoma by 3 rikishi is perhaps part of the reason why Chiyoshoma managed to only be separated by 2 rikishi - it's part of a compromise.

Edited by Gurowake

Share this post


Link to post
Share on other sites
32 minutes ago, Gurowake said:

There was a very good reason to separate them - just not the one they went with.  However, the fact that you thought they should all be next to each other while I thought Okinoumi should be split from Chiyoshoma by 3 rikishi is perhaps part of the reason why Chiyoshoma managed to only be separated by 2 rikishi - it's part of a compromise.

If I may ask, what were the reasons?

I wonder what the conventional wisdom is re the effect opponent rank/record has on the evaluation of one's own rank/record for the purposes of banzuke making - correct me if I'm wrong, but I think it's thought to only matter in edge/key cases like promotion, right? I'm wondering how much trouble I need to go to to prove my tinfoil hat theory.

Share this post


Link to post
Share on other sites
11 hours ago, Seiyashi said:

If I may ask, what were the reasons?

I'm doing this from memory so I might have the details wrong.

Aoiyama and Myogiryu couldn't be promoted, and there was no one else left to fill in the gap before they could be placed other than Okinoumi.  The latter should have been ranked behind the former two, but it simply wasn't possible.  After Okinoumi, there were those two 7-8s who could be slotted in, plus Nishikigi who was equivalent in rank/record to Chiyoshoma who should also be added.  That's how I got three in between them.   

Share this post


Link to post
Share on other sites
32 minutes ago, Gurowake said:

I'm doing this from memory so I might have the details wrong.

Aoiyama and Myogiryu couldn't be promoted, and there was no one else left to fill in the gap before they could be placed other than Okinoumi.  The latter should have been ranked behind the former two, but it simply wasn't possible.  After Okinoumi, there were those two 7-8s who could be slotted in, plus Nishikigi who was equivalent in rank/record to Chiyoshoma who should also be added.  That's how I got three in between them.   

Also, as I recall, trying the get the 5-10's closer together meant separating adjacent 9-6 Tochinoshin and Nishikigi further (which is what my doomed entry ended up doing I think).

Share this post


Link to post
Share on other sites
1 hour ago, Reonito said:

Also, as I recall, trying the get the 5-10's closer together meant separating adjacent 9-6 Tochinoshin and Nishikigi further (which is what my doomed entry ended up doing I think).

So basically, it's a case of screwing over everyone a little bit rather than stiffing someone hard for consistency elsewhere...

Share this post


Link to post
Share on other sites
27 minutes ago, Seiyashi said:

So basically, it's a case of screwing over everyone a little bit rather than stiffing someone hard for consistency elsewhere...

When there's a single most logical solution, they usually go with it. When there are multiple comparably logical, internally consistent solutions, they often end up with some inconsistent muddle in the middle.

Share this post


Link to post
Share on other sites

John Gunning actually said yesterday in his twitter spaces conversation that online banzuke aficionados probably put together more logical and consistent banzuke than the committee does, having both fewer biases and more intellectual firepower.

9:45 here: https://twitter.com/i/spaces/1vAxRkMABdrKl

  • Like 2

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, Seiyashi said:

So basically, it's a case of screwing over everyone a little bit rather than stiffing someone hard for consistency elsewhere...

Well, to spell it out Tochinoshin went +6 instead of +3, Okinoumi went -2.5 instead of -5,  Nishikigi went +5.5 instead of +3, Aoiyama went -0.5 instead of -1.0, Chiyoshoma went -3.5 instead of -5, Myogiryu went -1 (vs -1), and Sadanoumi went -4 instead of -5. So it begs the question of why they didn't go a little more lenient on Aoiyama and Myogiryu.

EDIT: it's also somewhat egregious as Aoiyama entered day 15 at 7-7 and lost to incoming Komosubi Daieisho to get dropped, whereas Myogiryu entered at 6-8 and beat the extremely-promoted Tochinoshin.

Edited by Hoshotakamoto

Share this post


Link to post
Share on other sites
7 minutes ago, Hoshotakamoto said:

Well, to spell it out Tochinoshin went +6 instead of +3, Okinoumi went -2.5 instead of -5,  Nishikigi went +5.5 instead of +3, Aoiyama went -0.5 instead of -1.0, Chiyoshoma went -3.5 instead of -5, Myogiryu went -1 (vs -1), and Sadanoumi went -4 instead of -5. So it begs the question of why they didn't go a little more lenient on Aoiyama and Myogiryu.

EDIT: it's also somewhat egregious as Aoiyama entered day 15 at 7-7 and lost to incoming Komosubi Daieisho to get dropped, whereas Myogiryu entered at 6-8 and beat the extremely-promoted Tochinoshin.

On some banzuke, they've gone out of their way to demote 7-8's, like this time, except for some reason Yutakayama. On others, including all recent ones, they've been perfectly happy to leave them in place. No idea what accounts for the difference.

  • Like 1

Share this post


Link to post
Share on other sites
10 minutes ago, Reonito said:

John Gunning actually said yesterday in his twitter spaces conversation that online banzuke aficionados probably put together more logical and consistent banzuke than the committee does, having both fewer biases and more intellectual firepower.

Out of reacts, but equal parts "thanks" and "haha".

So basically, the clue to hack GTB is, seat of the pants?

Share this post


Link to post
Share on other sites
Posted (edited)
4 minutes ago, Reonito said:

On some banzuke, they've gone out of their way to demote 7-8's, like this time, except for some reason Yutakayama. On others, including all recent ones, they've been perfectly happy to leave them in place. No idea what accounts for the difference.

Well clearly losing to a Komosubi is just unacceptable! (Laughing...)

EDIT: maybe the new strategy for sniffing out bias should be to look at who gets the most outrageous Day 15 matchup and then fade their chances to get a fair shake in the banzuke.

Edited by Hoshotakamoto

Share this post


Link to post
Share on other sites
5 minutes ago, Seiyashi said:

Out of reacts, but equal parts "thanks" and "haha".

So basically, the clue to hack GTB is, seat of the pants?

Someone suggested that I just feed all of the CSVs of 30 years of banzukes and records into Google's Vertex AI and pay them a few dollars to see how well they can score in GTB.

  • Like 2
  • Haha 1

Share this post


Link to post
Share on other sites
Posted (edited)
4 minutes ago, Hoshotakamoto said:

Someone suggested that I just feed all of the CSVs of 30 years of banzukes and records into Google's Vertex AI and pay them a few dollars to see how well they can score in GTB.

I was going to learn to write my own NN for that to be honest, but we're going to run up against the fact that banzuke making trends change from time to time. Some iterative parsing through a k-nearest neighbours algo might be necessary to see if we can get some AI insight into these trends, because I don't think it lines up neatly with 2-year trends like the makeup of the shimpan committee. Then afterwards you need to create sub models for each class, or something.

Edited by Seiyashi
  • Like 1

Share this post


Link to post
Share on other sites
Posted (edited)
3 hours ago, Reonito said:

John Gunning actually said yesterday in his twitter spaces conversation that online banzuke aficionados probably put together more logical and consistent banzuke than the committee does, having both fewer biases and more intellectual firepower.

Only marginally related, but yesterday I tested how the new banzuke would work out if I treated it as one for Seki-Toto (I used to do that regularly, but not in a while now), and other than the unavoidable crunch at M1/2, my standard set of ground rules would have been sufficient to create a valid ranking all the way to J14 without a single spot that required manual intervention. (Laughing...) That surprised even me, considering how awkward several areas of the real thing were when trying to do them Kyokai-style. (I hasten to add that I don't think it's a better ranking, since I'm accepting of a lot of things that would probably cause a ruckus in the real one. But it's definitely highly consistent.)

Edited by Asashosakari
  • Like 1

Share this post


Link to post
Share on other sites
6 hours ago, Hoshotakamoto said:

Someone suggested that I just feed all of the CSVs of 30 years of banzukes and records into Google's Vertex AI and pay them a few dollars to see how well they can score in GTB.

Just cue-balling here, but considering that even blackjack and roulette can be hacked, maybe the last truly fair game is GTB, since it is based on a truly quantum-level random system: the NSK banzuke committee.

  • Haha 1

Share this post


Link to post
Share on other sites
6 hours ago, Hoshotakamoto said:

Someone suggested that I just feed all of the CSVs of 30 years of banzukes and records into Google's Vertex AI and pay them a few dollars to see how well they can score in GTB.

I wouldn't bother gathering 30 years - some very important assumptions about promotions were discarded around 2005, and banzuke-making looks completely different ever since.

Share this post


Link to post
Share on other sites
23 hours ago, Seiyashi said:

Not sure that explains the distinction between the two Chiyos, does it? Both "scored" the same wins (albeit Chiyonokuni had withdrawals rather than losses for some of them) and both were dropped from the division from the same rank. In fact, the prevailing wisdom before the banzuke dropped would have squared with your theory, since most people would have kept Chiyomaru, but not now.

You might be on to something re Meisei and Ishiura, but if there's one thing the banzuke makers are consistent about, it's being annoyingly inconsistent with these over a long period of time, and sometimes even in the same banzuke - my personal grievance is with the separation of the 5-10 mid-maegashira trio when there was no apparent reason to do so. My working theory for that is that they put the KKs where they think they should go and the MKs fill the gaps, but no doubt someone will come and put holes in this too (for one I think there's too many MKs in the lower half of the banzuke to not need to bump some KKs up or down to make it work).

I suppose I didn't really explain what I was doing with the chiyos. Due to the same-rank-and-record situation I was treating them as a pair that had to fall very similarly. I think that this idea still meshes with what we have seen before with other "paired" demotions and promotions. I'm not sure if it meshes well with my new theory about the missed bouts given chiyonokuni's absences should maybe cause a larger drop? But I can definitely see an oyakata making the convincing argument that you can't keep chiyomaru without keeping chiyonokuni while another oyakata prevails in arguing that their results are different because of withdrawls.

Actually the more I think about it, the more it starts to make sense that vague principles alone guide the decisions on the banzuke. There's a handful of people presumably offering different interpretations and compromise is reached independently of a consistent system of internal notes.

Share this post


Link to post
Share on other sites
Posted (edited)

I've just realized that Sandanme has only 90 ranks this basho instead of 100. Has this happened before? I always thought that only the last two divisions had a non-fixed number of ranks

Edited by Leonishiki

Share this post


Link to post
Share on other sites
6 hours ago, Leonishiki said:

I've just realized that Sandanme has only 90 ranks this basho instead of 100. Has this happened before? I always thought that only the last two divisions had a non-fixed number of ranks

Divisions have changed size before. In this particular case, Shibatayama announced prior to the banzuke that due to lower intake numbers they would reduce the size of sandanme to keep things relatively proportional. 

  • Thanks 2

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now