Hoshotakamoto

I forgot to mention at the bottom of the previous post that I did get Aoiyama in exactly the right spot, but it was another Shirokuma situation where I had him behind Kotoeko and Oshoumi taking a heaping plate of disrespect which thankfully he was not served. So here's where we are heading into the seeding around the heaven and hell line. So at J12e they had Kotoeko grabbing that first spot (remember I had him at J11e) and then Tsushimanada claims J12w. No major surprise. I did mention that I had Oshoumi way too low, and this is because I thought Akua might be preserved at J12e. He was not. The J14 row seen above is a faithful rendition of the actual J13 line, and this is appropriate because we only have 26 juryos on this banzuke. Kitaharima does grab the J14e spot, which is no big surprise, and then if you thought Onokatsu was going to take J14w well then you're crazy, but if you thought Chiyosakae was going to take J14w then you're also crazy but this time you happen to be Right! Poor Chiyomaru has to stay at Ms1e (granted he did bring this upon himself with a record that calculates to Juryo 17 east). And if you heard me say "boy I won't be surprised if Onokatsu falls all the way to Ms2w" then you will be surprised to learn that I am not surprised because Onokatsu did in fact fall to Ms2w. Tenshoho and Yuma do claim the Ms1w and Ms2e (the order of who gets which cannot be in doubt).

Moving on down to row 8 - and I'll try to focus on leading with what the committee decided instead of leading with speculation and following up with answers. For J8 they went with the obvious choice of Asakoryu - Shishi, and on J9 Shimanoumi and Tamashoho were underdemoted, with 6-9 winning the tiebreaker vs 4-11, so it's Asakoryu - Shishi Shimanoumi - Tamashoho Shiden is still up next, and Wakatakakage is floating ... they decided to put Shiden in the J10e slot and let Wakatakakage get the J10w spot ahead of Oshoumi's juryo KK. Notice we're still waiting to learn about Aoiyama's parachute which definitely won't fall the whole way through the division. On the J11 line Oshoumi is awarded the next spot, right behind Wakatakakage, and finally J11w goes to Aoiyama - keeping him ahead of the non-Wakatakakage promotees from Makushita and the folks like Kotoeko and Akua who didn't need to finish out ahead of Aoiyama. *deep breath* now that that's all been said, how did the math predictions do .... let's see Asakoryu and Shishi was too easy, to the point of being impossible to mess up. On the next 2 rows however it looks like only 2 out of 8 points will be claimed, with Tamashoho / Shimanoumi reversed in the wrong direction and split across Wakatakakage (so Tamashoho costs 1 point, Waka costs 2 points, and Shimanoumi costs 2 points) and then Shiden is placed on the wrong side, well behind Wakatakakage, to drop 1 more point and salvage one. We finally arrive at the bottom 3 rows, with only 4 rikishi whose names appear on this goofy banzuke (makushitas were severely under promoted for the sake of illustration), so it will likely be those 4 and two wild cards (in fact it's obviously Kitaharima and a Chiyo). But let's save that for the next post...

So to summarize the last couple posts, here's the updated official banzuke down to J3. One would think at this point that Kagayaki and Oshoma are obvious picks, with the caveat that Bushozan is likely to catch a parachute ride. Ignoring Bushozan it would make sense to put Hakuyozan at J5e and leave Shirokuma at J5w. So what did they do? (checks notes) Well according to this they went with Kagayaki and Oshoma, and then the Bushozan parachute dropped him in to J5 (so no overpromotions for Shirokuma / Hakuyozan) - Shirokuma got his 1 slot promotion, and then Hakuyozan was underpromoted by half a slot. This means Chiyoshoma will get overdemoted. Does Chiyoshoma slide over to the next spot and disrupt those stacked KK winners? Yep he sure does. Does 10-5 trump 9-6? Yes, Hidenoumi claims J7e and Daishoho has to settle for J7w and the 2 row promotion on a 9-6. The sweet spot (recall Meisei vs Oho) gets another strike against it. Now let's tally up the predictions that were made from J4e down to J7w. Well Kagayaki and Oshoma were an easy call. My predictions on the "first iteration" were almost exactly what my intuition told me in the first paragraph, so (1) Bushozan was way out in left field (minus 2 points), (2) Shirokuma was a bulls eye ONLY because I had the wrong guy in front of him (2 wrongs make a right), (3) as I mentioned Hakuyozan was on the wrong row, and Chiyoshoma/Daishoho/Hidenoumi were all minus ones for reasons mentioned above. So in total that's a minus 5, or a 3 out of 8. Arguably the banzuke as a whole is a 3 out of 14, or maybe 5 out of 14 if there was another Shirokuma further up.

Juryo retrospective - seems that the screen shot I provided above is adequate to kick things off. I actually have no idea where anyone landed aside from Takarafuji at M1w so I'll be learning this as I go. Even though I liked Hokuseiho to stick around in division 1 for the math banzuke, it is a clear cut Tokihayate - Tohakuryu starting point. So naturally Tohakuryu's 8-7 finish got him a 0.5 slot improvement from J2w to J2e and Takarafuji was kept around for J1w which is a 2.5 slot drop in the sense that he was the 39th rikishi on the banzuke before and now he's the 44th rikishi on the banzuke. Hokuseiho gets passed over, amusingly, and Mitoryu is kept for J2w. Tomokaze, despite starting a full slot above Takarafuji, lands 2 rows below him on the 5 win vs 6 win outcomes, while Hokuseiho's ghost occupies J3e. So in a nutshell they kicked Hokuseiho down one slot, pushed Tomokaze over to the west side, and swapped Tohakuryu with Takarafuji because bias. If this had been a GTB scenario, then the "strictly following the math" outcome would have been docked 2 points for Hokuseiho, probably docked 2 points for having Mitoryu in his place, and would have lost 4 points on Tohakuryu / Takarafuji. And this is not counting the fact that my old math posts had Endo occupying the J1e slot (or that my "first iteration of thinking about Juryo" pushed Tokihayate down to J3). At least with the knowledge of which 42 they kept in makuuchi, it was a bit more "obvious" to put Tokihayate in the first available slot. And knowing Hokuseiho was a non entity could have helped some as well.

So pretty much it's just Endo and Takerufuji walking in first on the east side like

It looks like one of them may have been aided by what became the scandal.

Let's put a bow on the Makuuchi portion of this retrospective: Kitanowaka sure looks like an easy pick for M14w. And indeed, that pick is cashed in across the board. This is where almost all of my fun ends. We have a mathematical tie across the board for J1 Daiamami, J3 Roga, and J10 Takerufuji. I argued publicly that Takerufuji is the most deserving - and maybe he is - but no one saw it that way and I didn't even send it off that way. I sent off Daiamami - Roga to GTB and let Myogiryu ahead of Takerufuji (ahead of Endo) and they didn't see it that way either. They decided to let Myogiryu - almost demoted for having a 5-10 at M12w when there is no M17w - jump the line ahead of these three and claim M15e. Well ok then. They then went ahead and randomly let Roga jump ahead of Daiamami but also stay ahead of Takerufuji, which I can almost agree with except that not only did they do that but they pulled Endo along for the ride, submerging Daiamami and Takerufuji under the pile. I can understand Takerufuji coming last place but the rest of this feels totally random. I scored the point for having Roga in the right spot for GTB, because I had him in 2nd place out of these five jokers, but only 2nd to Daiamami, not 2nd to Myogiryu and ahead of Daiamami and the rest. And then of course I let Hokuseiho in instead of Endo for the public math post, which doesn't warrant any further discussion. Final score for math-the-banzuke: 42 out of 84 (50% - the result of many coin flips?). Final score for my official submission: 43 - thanks Roga? Final score for correctly keeping in Endo, Takerufuji, and Myogiryu while keeping out Hokuseiho, Tokihayate, and Tohakuryu: zero. It doesn't always pay to be correct. Stay tuned for juryo analysis on a later date when my wife isn't ready to kill me.

Ok we're heading into the home stretch of Makuuchi. I'm debating if I've made this too confusing by always putting the graphics below the text. Let's try putting analysis under the graphic. So for M11w our choices look like: (1) break a tie between Sadanoumi and Shonannoumi and put the winner in M11w, (2) Give Shimazuumi the go-ahead for KK (from 5.5 slots further down the banzuke, hmmm), or (3) give Ryuden a nice parachute. The official decision was to let Sadanoumi the veteran win the tiebreaker and drop 1 spot for 6-9. The decision I posted publicly was to give Shonannoumi the benefit of dropping in from M6w, and the decision I made for GTB was to give Ryuden the parachute and land him here (actually I had Shonannoumi's parachute landing him above all of this and Mitakeumi getting the shaft). I still sortof believe in my Shonannoumi / Ryuden choice, but I guess the takeaway here is that they were willing to go with Mitakeumi and the math this time around instead of letting these guys land softly. But now the choice is deferred to M12e - so which of the above do we believe in? The committee decided to let Shonannoumi go next since the math supports him above Ryuden and Shimazuumi. Certainly defensible. The impact is that my math post had the right idea but broke the tie wrong, so minus 4 points. My GTB entry gets no soup, for putting Mitakeumi in M12e. And now the committee goes with M17e Shimazuumi winning the math-tie over M5e Ryuden which I guess means that they hate Ryuden (also defensible). This actually worked out for my GTB entry (having Shimazuumi in exactly the right slot), but only because I was erratic and put Sadanoumi underneath all of them. Once this bit of craziness is resolved, it is fairly academic to slot in Churanoumi (the 7-8 maegashira) over Nishikifuji (the 10-5 juryo) at M13w/M14e and I got full credit across the board for these two picks, as did I'm sure just about everyone. Available points down to M14e - Nishifuji: 72 MTB points: 40 (only cashing in the last 2 easy picks), my GTB score: 39 (those last 2 plus Shimazuumi emerging from the chaos I created).

It's not hard to make sense out of the next few official decisions, given what has happened so far. They do go with Onosho in M8e, and finally Takayasu drops in at M8w, ahead of Hokutofuji, indicating 0.5 slots of sanyaku bias (no big deal - although it is a big deal for my MTB entry since it knocks 4 more points off). They went with Kotoshoho (M9w) over Shodai (M10e) which was supported by the math although any expectation of joi bias was dashed in the process. My choices in the official GTB submission to (1) move Takayasu way up to M4, (2) drop Onosho below Hokutofuji, and (3) swap Kotoshoho with Shodai meant I scored 0/8 for M8/M9 (I also lost all the points at M10). When it came time to break the tie for M10 Mitakeumi's 6-9 (and ozeki past?) prevailed over Ichiyamamoto's 5-10 from a higher spot. I got this wrong in both banzukes where I moved Ichiyamamoto ahead, and did myself no favors by dropping Mitakeumi to M12e in my official GTB entry. Counting up points through M11e - Ichiyamamoto I get Available points: 60 MTB points: 36. My official GTB points: 33

to think I could have just said 12-3 and gone from very wrong to oh so right

So here's the new lay of the land: Onosato has fully lapped 9-6 Tsurugisho who had a 4.5 row advantage (maybe that sweet spot wasn't SO sweet). Surely Kinbozan has to uphold his M6e now with the 7-8. Gonoyama is almost certain to be mistreated in comparison to Midorifuji. Takayasu still looms from Komusubi. From my MTB entry I have Tsurugisho > Kinbozan > Gonoyama, with Tsurugisho's name on the M5 row. From my official GTB entry I have Kinbozan > Gonoyama > Tsurugisho, with Tsurugisho's name on the M7e row. The Gonoyama pick at M6w cashes in 2 points (this made more logical sense to me with Midorifuji directly above him at M5w), but Tsurugisho at M6e is a total loss for me, and Kinbozan falling to M7e behind Tsurugisho, Onosato, and for that matter Gonoyama just seems petty. At M7w Tamawashi is an easy pick, which was documented in my MTB post, but of course he landed at M8e in my GTB entry since I had Takayasu above all of these people. Available points: 46 MTB points: 30. My official GTB points: 33

Now is where things decidedly do not go my way. We have to put someone's name in M4w. It won't be Kinbozan. The next favorite is Midorifuji - do we drop him 2.5 rows for his 5-10? My thinking on review was that Takayasu might be the wild card here to fill a gaping chasm in the banzuke. If memory serves they basically did agree with my idea to use him exactly this way a couple GTBs ago. Even though Midorifuji calculates to M7e and Takayasu calculates to M11e I thought maybe they'll forgive the 2-4-9 over the 5-10. But no one agreed with me, least of all the ones making the decisions. I had Midorifuji M4w and Hiradoumi M5e when I posted publicly. The official decision was to elevate Hiradoumi's 8-7 above Midorifuji's 5-10 and give Hiradoumi the full bump of 4 rows for his 8-7 finish. Because Midorifuji falls to M5e and I had kicked him down to M5w I salvage 1 point in my official GTB submission but the rest of this is a mess. Now what does this leave? We still need to fill M5w, and the top person available is Kinbozan. He is not moving up. Now we have 5-10 Gonoyama available from M3e and M11e Tsurugisho from 9-6. So of course they officially run with the sexy pick and elevate 11-4 Onosato 10 rows from M15w. Note that this is almost exactly the opposite of anything we've seen in the past few years. I'll pause with M5w filled. Available points: 38 MTB points: 26. My official GTB points: 31 Shots fired, by the committee.

Now I'll zoom in on my Takayasu debacle. Reading the stacks below, it seems clear that Tobizaru will hold up as M4e with his 7-8. He mathematically has the best claim to M2w, but he will not be promoted above his starting position. Meisei vs Oho is the first place where I changed my mind. First I agreed with the popular position, which was that Oho's 10-5 would put him at M2w and Meisei would take M3e. When the time came to submit my entry, I decided 9-6 was the "sweet spot" where you don't get overtaken by 10-5, and whether or not this argument holds water, it did turn out that they decided to agree with me this time. Now we have to fill M3w. It seems clear that it will be Midorifuji or Takanosho, since they are part of a 3 way tie and Kinbozan isn't going up. I also decided to keep Takayasu's name in play - even though he calculates to M11e, his Komusubi starting point has its advantages and the former Ozeki is a banzuke favorite. The popular decision here was to go with Takanosho's 10-5, it was the one I posted for math-the-banzuke, it was in my GTB submission, and it was the "correct" choice. Available points: 30 MTB points: 24. My official GTB points: 28 (I jumped on board with Meisei for fun and profit).

All right let's do a math retrospective and see what the decision makers thought of the banzuke math. I revert to the unstacked changes, although Abi and company move up because there wasn't a lot of room for debate above that line. My first observation is that Nishikigi moves up to K1w (most people knew this) as the most eligible claimant of K1w. Ura could not hold his spot and Nishikigi calculated ahead of the rest. Subsequently the committee went the route of counting Ura's 6-9 as -0.5 instead of moving Asanoyama ahead of him. This cost me 2 points in my MTB and my official GTB. The official decision also promotes consistency in the treatment of Ura and Atamifuji, which I might have given more consideration to. Before I move on, I'll also note that Atamifuji is the obvious pick for M2e, and they agreed. Available points: 24 MTB points: 22. My official GTB points: 22 (no disagreement yet vs what I had posted publicly).

It seems obvious that the 57 point entries are going to be top 5 material.

If they were going to destroy everyone's GTB entries no matter what, why did they not pull the trigger on sparing Takarafuji for another run at Makuuchi...

I'm reading my official submission and the version I posted above landing around 41 and 42 a piece, with my updated version jumping out to an early lead (starting from the top down) and then being overtaken by the middle of the banzuke. I'll do a more indepth look at the choices later.

placeholder for natsu 2000 real basho (not sure if i will ever actually get to these)

Prepared the meta banzuke for Natsu 2000, based on results from http://sumodb.sumogames.de/gtb/GTBScoreBasho.aspx?b=200005. First things first, Sekiwake Ekigozan had a 7-8 and dropped to Komusubi, so he basically needs a new ozeki run despite 37 wins to get from M14 -> K -> S -> K. But we did have a makuuchi yusho winner as Imanonami pulled 14-1 and jumped to M1e. There was a weak case for K2 but M1 was about right when reviewing all relevant precedent in sumo. So with a K1e and a M1e we have opponents for the mythical yokozuna and ozeki to face in the first week. Mediocre results for the two Juryos had the aggregate impact of a slight demotion, but Ms2 Leonishiki pulled 5-2 to join them at J11w. Elsewhere the ranks in Makushita would have thinned out considerably but Gaijingai's 15-0 as the overall GTB winner lifted him from Sd10 to Ms12, just ahead of the spot vacated by Doitsuyama's kyujo. Sandanme will have a solid group of 10 eligible to compete next time, but here I'll note that I failed to mark Okimura intai at Sd47 (the highest ranked departure yet) in the previous basho, and likewise for Jd9 Nikkan Sports. Jonidan will have 12 eligible competitors and there were 8 newcomers that may or may not return for the next Jonokuchi bracket. I did make an attempt to thin out the unused rows in the right side but it didn't go exactly as I planned. I'll make a more aggressive push for the next banzuke to focus on the essential rows and leave out the giant gaps in divisions 4-6.

placeholder for later discussion of the proper Haru 2000 basho

Here's the banzuke for Haru 2000, highlighted with Ekigozan's back to back 15-0 finish, moving up from K1w to S1e (and it's fair to say any intermediate ranking would have gotten him to S1e from M14 with the 30-0 run). Is this an ozeki run? Hard to say. I can't fathom a scenario in real life where a rikishi could enter at M14, go 15-0 to Komusubi, 15-0 to sekiwake, and then be held back from Ozeki if they completed a 45-0 run. Meanwhile we had stellar finishes for both Juryo contestants, and Imanonami only dropped a couple spots with 6-9. In Makushita we didn't have a runaway 7-0 (usually a 14-1 or better, but sometimes a 13-2 is good enough) but two 6-1 finishes brought Leonishiki and Doitsuyama within one 7-0 finish of Juryo. After detailing my whole process I forgot to mark any intais, and it would require me to re-seed the whole bracket if I fill them in now. I'll clean this up in the next round. I will say it is apparent that the recruiting boom starts with this basho, and advances in scripting and automation are essential in order if I want to avoid this thing dragging out forever.

It would be kinda neat when the cards are all on the table to have a simulation that would demonstrate how fates would have diverged if slightly different promotion criteria were used (very generous vs very harsh). I'm telling myself it doesn't make a very big difference at this point since people generally leave (early on) without leaving a big footprint.

The manual patterns that I established the past couple days were to (1) refresh the rows that are available to populate with rikishi, (2) lookup the rank where they should go based on past historical patterns and internal consistency with conclusions I've reached already, (3) reload the hacked version of chiyotasuke's tool and move everyone into their new spot, (4) post the snapshots to imgur and write a new forum post, (5) zero out everyone's record to 0-0-7 for the next basho and manually type in their new landing rank, (6) run a script that would read the scoreboard for the next GTB and populate their new records, (7) manually look up who went intai and write their names or erase their names as needed, and I think we're roughly at the point where the process repeats. The place where I'd stub my toe is on step 3 if I hadn't properly done step 1 to make sure the rows where there that i wanted, so a bit of bookkeeping went in place to make sure I didn't have to keep stopping and starting here over and over. I'll probably automate 1 today as I discussed in the earlier post, (2) is the script that I mentioned above so this bookkeeping is a lot better than it has been. (3) i'll put off automating as long as possible since it's the fun part (but I might automate it if it stops being fun), (4) will always be manual but it only takes a couple minutes, (5) is easy to automate, (6) i'll probably automate today, (7) same thing. I know this doesn't succinctly answer your question but it paints the big picture.

I don't know, I wrote a script last night (GPT kinda wrote a lot of it) to prompt me to keep proposed ending ranks fresh in a file when I don't have them from earlier runs. I will probably tweak Chiyotasuke's page to automatically show me the rows on the banzuke that I care about and churn some of the big empty sections that are accumulating. script.js mentioned below is literally Chiyotasuke's script.js where the rikishi are listed but index.html is where his page hardcodes the rows for the left banzuke and the right banzuke (although if you save your board and reload it the InnerHTML for the actual banzuke will be refreshed from the cache, which is something that slows me down :) )

Thank you. I did notice last night that Jasonishiki changed and I wondered if I had missed any. Since sumodb kept going down last night I cached local copies of the first few dozen rikishi so I wouldn't have to pause my efforts during later outages. Also wrote more scripting to speed along the process as the rosters get bigger and bigger.