Sign in to follow this  
Kintamayama

GTB invitation- March 2024 - Results!

Recommended Posts

3 hours ago, Hoshotakamoto said:

I was thinking Yokozunas would be put out to pasture but they could come back at Ms10, like ascending in Cookie Clicker.

What would be the demotion threshold though?

Share this post


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

I was thinking Yokozunas would be put out to pasture but they could come back at Ms10, like ascending in Cookie Clicker.

As dishonored Yokozunae will still be on  the "other" banzuke, I think maybe treat them as super - Ozeki demotion-wise (three consecutive MKs?) , simply decide that Yokozunae are demotable. Decide on number of wins they need to be immediately re-promoted, and let them slither down the banzuke honorably if needed. Like in Sumo game.

Edited by Kintamayama
  • Like 1

Share this post


Link to post
Share on other sites
On 15/02/2024 at 03:46, Hoshotakamoto said:

ah ... well now this is going to happen, based on results since the beginning of time.

As I hinted at earlier, I already gave it a shot, just as a trial to learn things that would need to be addressed for more a thorough attempt. A few topics that warranted consideration / things I learned from the process.

- Size of each division. Given the ups and downs in player count over time, proportional division sizes seemed most logical.

- How to score each of these divisions? The players in the top divisions will get there because they consistently score KK (in traditional GTB scoring). You can't have everyone in the top division get KK, so you need to pit them against each other. How to score wins and losses though? My leading idea was (for 15-day records)... 'Player's points' minus 'average points scored in the division' divided by '1.5 times the standard deviation of points in the division', plugging that into a NORMSDIST function, to get a number between 0 and 1, then multiplying that by 15 to get the number of wins. Round down all KK, round up all MK (and if win-loss ratio for the division is not equal, round up the guy that computes 9.99 wins to 10, or round down the guy that computes 5.01 wins to 5). That produced a decent distribution of scores, and the yusho line was usually 13-2. I didn't really think too much through a 7-day scoring system, probably something along similar lines except plot the NORMSDIST result against the actual disbribution of scores in that division in sumo (actually you could probably do that for 15-day scoring too, I just haven't tested it thoroughly).  

- How to compute movements in the divisions? Gurowake posted a chart not terribly long ago on typical or average lower division movements. As is known, promotions tend to get smaller the closer to the top of a division, interestingly so, it does so in a fashion that reasonably approximates a linear regression. That means it can be plugged in easily to fluctuating division sizes, so that players will always move proportionately in the division, no matter it's size. So for example, the following equation for a Sd rikishi with 5 wins is (assuming a Sandanme that goes to 100)

y = X+(-0.15305*X+0.41574)

Where Y is the computed placement, and X is the current placement.

Test run... Sd50 (X = 4.49. 4 = division, 0.49 = proportional placement in division (Sd1 = 4.0)). Y = 4.218, which would be Sd 22/23. Since 1998 (the start of GTB), 5 wins at Sd50 puts you between Sd19-26

Sd25 computes to Sd1, actual result = ms60

Sd100 computes to Sd65-66, actual result = Sd60-71

- Yokozuna: Given that GTB is more of a crapshoot than sumo, back to back yusho would be extremely rare (though I never really ran into a shortage of Ozeki). Demotion criteria, I was treating them like super-ozeki. 10 wins to avoid kadoban > fail to clear kadoban > demotion to 'Yokozeki' > 12+ wins > back to Yokozuna.

- Automation... it is quite time consuming to do by hand (not so bad for each individual banzuke, but we have 25 years to catch up on). I know a lot of people here have data management experience, so I'm sure they could come up with something quite efficient. 

  • Like 1

Share this post


Link to post
Share on other sites
On 15/02/2024 at 09:46, Hoshotakamoto said:

Here's the end of the Haru 1998 contest (not in the order that they finished, since stronger rikishi with the same record often start and finish behind their counterparts in Jk)

...so on what basis did you order them?

Share this post


Link to post
Share on other sites
3 minutes ago, Asashosakari said:

...so on what basis did you order them?

What I think I've witnessed in Jonokuchi is that the strongest contenders out of Mz are ranked in backwards order so if there's a 7-0, 7-0 tie then the juggernaut often lands 1/2 slot behind the runner-up at Jd15e/w. So for the 6 people I ranked instead of 1-2-3-4-5-6 I think they landed something like 1-2-4-3-6-5 with tiebreakers going the wrong direction for the 4 win and 2 win finishers (this is from memory, not reviewing my post above).

Share this post


Link to post
Share on other sites

I need to clear the NaNs on the next iteration but here was the outcome of the 2nd GTB basho. I'll give these their own thread once the machine is oiled and running.

deGHlYD.png

Share this post


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

What I think I've witnessed in Jonokuchi is that the strongest contenders out of Mz are ranked in backwards order

My goodness. If this is what happens when people try to make sense of maezumo on their own, I must whole-heartedly apologize to the forum for slacking on my summaries since 2021. No, that's not how it works whatsoever.

Dare I ask what other banzuke misunderstandings are going to find their way into this project?

Edited by Asashosakari

Share this post


Link to post
Share on other sites
14 minutes ago, Asashosakari said:

My goodness. If this is what happens when people try to make sense of maezumo on their own, I must whole-heartedly apologize to the forum for slacking on my summaries since 2021. No, that's not how it works whatsoever.

Dare I ask what other banzuke misunderstandings are going to find their way into this project?

A 15-0 record won't actually mean that a rikishi won 15 head-to-head bouts against other rikishi.

Share this post


Link to post
Share on other sites

Could someone please move the "GTB banzuke" discussion to the meta GTB thread?

  • Thanks 1

Share this post


Link to post
Share on other sites
12 hours ago, Asashosakari said:

 I must whole-heartedly apologize to the forum for slacking on my summaries since 2021.

I do have a legitimate question (and don't feel compelled to move the question to meta GTB) - what would typically happen if say a Kototebakari comes in so they are going to seed him, and he's opposite a veteran who fell off the banzuke and is fighting their way back? Who's more likely to be seeded higher / what factors go into it?

Share this post


Link to post
Share on other sites

On a related note is anyone else a little surprised to see these other 3 clustered as close to Koto as they are about to be on the next banzuke?

jpJbpEi.png

Share this post


Link to post
Share on other sites

GTB deadline has passed. Time to drop the script

 

Y1e 13-2 Y Y1        
        Y2        
O1e 11-4 O1 O1w 10-4-1
O2w 2-2-11 +0.5 O2 S1e 13-2 DG  ↑ 
        O3        
S1w 9-6 +0.5 S1 M1e 10-5 S  ↑ 
        S2        
        S3        
M2w
Abi
8-7  ↑  K1 M7w 9-3-3  ↑ 
        K2        
        K3        
K1w
Ura
6-9  ↓  M1 M5w 8-7 +4
M1w 6-9 -0.5 M2 M11w
Oho
10-5 +9
M9w 9-6 +6.5 M3 M12e 10-5 +8.5
M4e 7-8 M4 M2e 5-10 -2.5
M8w 8-7 +3.5 M5 M11e 9-6 +5.5
M6e 7-8 M6 M3e 5-10 -3.5
M15w 11-4 K +8.5 M7 M10e 8-7 +2.5
M14w 10-5 +6.5 M8 K1e 2-4-9  ↓ 
M3w 4-5-6 -5.5 M9 M14e 9-6 +4.5
M4w 4-11 -5.5 M10 M9e 6-9 -1.5
M7e 5-10 -4 M11 M10w 6-9 -1
M6w 4-11 -5.5 M12 M17e 9-6 +4.5
M5e 3-12 -8 M13 M13e 7-8 -0.5
J2e 10-5  ↑  M14 J3w 10-5  ↑ 
M12w 5-10 -2.5 M15 J10e 13-2 Y  ↑ 
J3e 9-6  ↑  M16 J1e 8-7  ↑ 
M13w 5-10 -3.5 M17
  • Like 1

Share this post


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

well if they drop Takayasu this far my fate is sealed

When I checked the history for 2 win komusubi I saw the second to last time it happened in 2014 the komusubi Takakaze was placed at 9w behind Tamawashi who got 8 wins at M10w, half a rank lower than he was this time with the same record. If they put Tamawashi ahead back then I figured these days with a lower amount of sanyaku bias Takayasu should at least fall to M8E and then whether he's put below Onosho or Hokutofuji is more debatable.

Edited by #1HENKAFAN
  • Like 2

Share this post


Link to post
Share on other sites
3 hours ago, #1HENKAFAN said:

When I checked the history for 2 win komusubi I saw the second to last time it happened in 2014 the komusubi Takakaze was placed at 9w behind Tamawashi who got 8 wins at M10w, half a rank lower than he was this time with the same record. If they put Tamawashi ahead back then I figured these days with a lower amount of sanyaku bias Takayasu should at least fall to M8E and then whether he's put below Onosho or Hokutofuji is more debatable.

I've got Takayasu in the same spot; he could go as high as 7e but I'd be surprised to see him any higher. Our entries are generally very similar; I went with Nishikigi at K instead of Asa, Shimazuumi and Ryuden the other way round (I debated that one for a while) and your M15 and M16 are my M16 and M15. Here's my guess.

Edited by Reonito
  • Like 1

Share this post


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

On a related note is anyone else a little surprised to see these other 3 clustered as close to Koto as they are about to be on the next banzuke?


Not particularly. Kototebakari and Hananoumi both have 56 wins, Wakanosho 53, and Daiseizan has no make-koshi yet - he would likely be ahead of them all if he hadn't missed two basho with, a broken hand I think it was. As ever, upper Makushita will be the real proving grounds, although Wakanosho was make-koshi in January so he won't be there just yet.

  • Like 1

Share this post


Link to post
Share on other sites

Could be anything from glorious victory to double-digit makekoshi:

Terunofuji (Ye 13-2 Y)   Y    ---
Kirishima (O1e 11-4)     O1   Hoshoryu (O1w 10-4-1)
Takakeisho (O2w 2-2-11)  O2   Kotonowaka (Se 13-2)
Daieisho (Sw 9-6)        S    Wakamotoharu (M1e 10-5)
Abi (M2w 8-7)            K    Asanoyama (M7w 9-3-3)

Ura (Kw 6-9)             M1   Nishikigi (M5w 8-7)
Atamifuji (M1w 6-9)      M2   Oho (M11w 10-5)
Meisei (M9w 9-6)         M3   Takanosho (M12e 10-5)
Tobizaru (M4e 7-8)       M4   Midorifuji (M2e 5-10)
Hiradoumi (M8w 8-7)      M5   Gonoyama (M3e 5-10)
Kinbozan (M6e 7-8)       M6   Onosato (M15w 11-4)
Tsurugisho (M11e 9-6)    M7   Onosho (M14w 10-5)
Tamawashi (M10e 8-7)     M8   Takayasu (Ke 2-4-9)
Hokutofuji (M3w 4-5-6)   M9   Shodai (M4w 4-11)
Kotoshoho (M14e 9-6)     M10  Mitakeumi (M9e 6-9)
Ichiyamamoto (M7e 5-10)  M11  Ryuden (M5e 3-12)
Sadanoumi (M10w 6-9)     M12  Shonannoumi (M6w 4-11)
Shimazuumi (M17e 9-6)    M13  Churanoumi (M13e 7-8)
Nishikifuji (J2e 10-5)   M14  Kitanowaka (J3w 10-5)
Roga (J3e 9-6)           M15  Daiamami (J1e 8-7)
Myogiryu (M12w 5-10)     M16  Endo (M13w 5-10)
Takarafuji (M16e 6-9)    M17  ---

Share this post


Link to post
Share on other sites
18 minutes ago, Asashosakari said:

Could be anything from glorious victory to double-digit makekoshi:

I don't think you have to worry about the latter.

Share this post


Link to post
Share on other sites

Normally I'm not eager to disagree with the GOAT banzuke guesser but according to the kaiju Takerufuji got his promotion

 

1 hour ago, Akinomaki said:

Terunofuji was interviewed the other day by a Sunday morning show on NTV, on air yesterday - the video was gone by today, I haven't found a proper summary, this is the only useful one http://www.moezine.com/1051471/

1708280585_maxresdefault.jpg

He properly explained the injury: the broken/crushed bone at the hip was a vertebra and what was squashed was an intervertebral disc

And very interesting: he presented Takerufuji as makuuchi next basho, not published yet - he himself had scouted him at Tottori Johoku high

 

Share this post


Link to post
Share on other sites
3 minutes ago, #1HENKAFAN said:

Normally I'm not eager to disagree with the GOAT banzuke guesser but according to the kaiju Takerufuji got his promotion

Even without seeing this, he has a clear promotion case and the guys he's competing with clearly missed their safety targets; I don't recall recent cases where the decision went against that.

Share this post


Link to post
Share on other sites
31 minutes ago, #1HENKAFAN said:

Normally I'm not eager to disagree with the GOAT banzuke guesser but according to the kaiju Takerufuji got his promotion

My whole guess is very hunch-heavy this time, hence also my expectation that it could all backfire spectacularly. If this is the only thing that's wrong, I'll be very happy.

FWIW, I treated it as an Isegahama-for-Isegahama question, and predictably went with the veteran for the last spot. As a fan, I'd prefer Takerufuji to get the spot.

Share this post


Link to post
Share on other sites
7 hours ago, #1HENKAFAN said:

When I checked the history for 2 win komusubi I saw the second to last time it happened in 2014 the komusubi Takakaze was placed at 9w behind Tamawashi who got 8 wins at M10w, half a rank lower than he was this time with the same record. If they put Tamawashi ahead back then I figured these days with a lower amount of sanyaku bias Takayasu should at least fall to M8E and then whether he's put below Onosho or Hokutofuji is more debatable.

I'm saying this because of how this banzuke is working out, not because I'm sure I did well: history is probably not going to be the most useful way of figuring out what they'll do this time. There are just too many gaps to fill, to the point where it would be more surprising if we didn't see at least a few moves that are at least tied for the biggest overpromotion/underdemotion for a particular record. (Maybe with the exception of extreme circumstances like scandals that force a bunch of kyujos/retirements.)

Share this post


Link to post
Share on other sites
Terunofuji(13-2 Ye) Y  
Kirishima(11-4 Oe) O Hoshoryu(10-4 Ow)
Takakeisho(2-2 Ow) O Kotonowaka(13-2 Se)
Daieisho(9-6 Sw) S Wakamotoharu(10-5 M1e)
Abi(8-7 M2w) K Asanoyama(9-3 M7w)
     
Ura(6-9 Kw) M1 Nishikigi(8-7 M5w)
Atamifuji(6-9 M1w) M2 Oho(10-5 M11w)
Meisei(9-6 M9w) M3 Midorifuji(5-10 M2e)
Tobizaru(7-8 M4e) M4 Takanosho(10-5 M12e)
Hiradoumi(8-7 M8w) M5 Gonoyama(5-10 M3e)
Kinbozan(7-8 M6e) M6 Tsurugisho(9-6 M11e)
Onosato(11-4 M15w) M7 Takayasu(2-4 Ke)
Tamawashi(8-7 M10e) M8 Onosho(10-5 M14w)
Hokutofuji(4-5 M3w) M9 Kotoshoho(9-6 M14e)
Shodai(4-11 M4w) M10 Ichiyamamoto(5-10 M7e)
Mitakeumi(6-9 M9e) M11 Ryuden(3-12 M5e)
Shonannoumi(4-11 M6w) M12 Sadanoumi(6-9 M10w)
Shimazuumi(9-6 M17e) M13 Churanoumi(7-8 M13e)
Nishikifuji(10-5 J2e) M14 Kitanowaka(10-5 J3w)
Takerufuji(13-2 J10e) M15 Myogiryu(5-10 M12w)
Hokuseiho(2-4 M8e) M16 Roga(9-6 J3e)
Daiamami(8-7 J1e) M17  

Share this post


Link to post
Share on other sites
15 minutes ago, Sumo Spiffy said:

I'm saying this because of how this banzuke is working out, not because I'm sure I did well: history is probably not going to be the most useful way of figuring out what they'll do this time. There are just too many gaps to fill, to the point where it would be more surprising if we didn't see at least a few moves that are at least tied for the biggest overpromotion/underdemotion for a particular record. (Maybe with the exception of extreme circumstances like scandals that force a bunch of kyujos/retirements.)

Just filling Kw-M3 will take us into near-record territory, although I'd argue history is still a useful guide to how they approach numerical ties or near-ties in various scenarios, even if it's not the exact rank/record/placement situations we're looking at here.

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
Sign in to follow this