GTB invite- Aki 2022 - RESULTS!!

[Sumo Spiffy 207]

2 minutes ago, Jakusotsu said:

It came together surprisingly easy once you applied a certain method to this madness, but of course it's far from certain that it's the correct method and the correct madness.

I'll talk about the specific methods I applied in my GTB video for this basho, but suffice to say two things:

1. I said I was going to wait in case more information came out, but I've also ended up waiting because there are a handful of ranks I've shuffled around quite a bit as I thought about what makes the most sense for the approach I'm using.

2. The video is going to have a lot of caveats. 

[Seiyashi 3,361]

1 hour ago, Jakusotsu said:

It came together surprisingly easy once you applied a certain method to this madness, but of course it's far from certain that it's the correct method and the correct madness.

Bold to assume that there is a method to the madness.  Does in vino veritas count as a method, or the madness?

[Asashosakari 16,827]

6 hours ago, Jakusotsu said:

It came together surprisingly easy once you applied a certain method to this madness, but of course it's far from certain that it's the correct method and the correct madness.

I had the same impression. As earlier comments alluded to, Ichiyamamoto's record is probably the biggest wildcard in how they treat it. (I suspect he's way too high where I've currently slotted him in.)

[Oshirokita 153]

I applied what seemed a consistent pattern to the withdrawn bouts and submitted a guess. Still seems a bit wonky. My biggest issue was how much to promote the Juryo promotee(s).

[Athenayama 455]

Isn't there a "cheat code" to achieve 15-0 at GTB? Where can I find it?

[Jakusotsu 4,878]

Depends on what the gyoji currently writing the banzuke is susceptible to.

[Athenayama 455]

I was only joking but I'm surprised how your answer had a serious tone.

[Jakusotsu 4,878]

Hard to be joking about serious matters.

[Oortael 131]

4 hours ago, Athenayama said:

Isn't there a "cheat code" to achieve 15-0 at GTB? Where can I find it?

If I'm not mistaken, nobody ever has scored the perfect 84 points in GTB. So no 15-0 zensho.
As some forum members have offered in the past, I'll eat my hat if it does happen in this particular basho.

2 hours ago, Jakusotsu said:

Depends on what the gyoji currently writing the banzuke is susceptible to inebriated with.

FTFY

[Jakusotsu 4,878]

21 minutes ago, Oortael said:

If I'm not mistaken, nobody ever has scored the perfect 84 points in GTB. So no 15-0 zensho.

And if I'm not mistaken it would have happened once, prompting disqualification and advancing the deadline due to obvious leaks.

 

23 minutes ago, Oortael said:

2 hours ago, Jakusotsu said:

Depends on what the gyoji currently writing the banzuke is susceptible to inebriated with.

FTFY

I'm pretty sure he would be sober by now.

[WAKATAKE 2,357]

Not sure which is harder, 42 perfect makuuchi placements or 63 perfect game winners in March Madness Tourney Pick’em

[Yarimotsu 208]

16 hours ago, Oortael said:

If I'm not mistaken, nobody ever has scored the perfect 84 points in GTB. So no 15-0 zensho.
As some forum members have offered in the past, I'll eat my hat if it does happen in this particular basho.

FTFY

I would love to know how statistically unlikely that actually is. It's got to be a fairly complex bayesian calculation to figure it out correctly.

[Seiyashi 3,361]

2 hours ago, Yarimotsu said:

I would love to know how statistically unlikely that actually is. It's got to be a fairly complex bayesian calculation to figure it out correctly.

Not that bad - Tedious, but not necessarily complicated, if you make the simplifying assumption that a rikishi has an equal chance of being ranked anywhere within the range of ranks that his rank and record combination would warrant based on historical data. Tachiai had an article a while back that collated this info. 

Where it gets complicated is where you try and add in rules, control for rikishi occupying the same rank, disregard certain impossible movements... Etcetc. Then of course there's all the covid madness for this GTB. 

The other way to go about it is to check the rarity drop off for the top results in GTB and extrapolate from there. If a 82 is so and so rare compared to 80 then a 84 must be much rarer than an 82 or something like that, so we could probabilistically expect a 84 in how many entries more. But even that also has vulnerabilities because not every entry is a meaningful possible entry. 

So the calculation isn't necessarily complex, but adjustments for the assumptions are, which might ne what you meant. 

Edited August 12, 2022 by Seiyashi

[Jakusotsu 4,878]

You'd have to calculate the odds for each rank indiviually. The placement of Yokozuna and Ozeki is predetermined, and for Sekiwake/Komusubi you have much more limited number of options than for Maegashira X.

Perhaps calculating the exact probability is even more unlikely than hitting the perfect 84 points.

[Asashosakari 16,827]

Am I the only one confused by this tangent considering the original wish was for a 15-0 record, which happens in every GTB edition?

[Seiyashi 3,361]

Whoops. Think we were discussing 84/84, rather than 15-0 (the two got conflated).

[Muhomatsu 136]

entered... although I have very little confidence in what I put down.

[Gurowake 3,452]

20 hours ago, WAKATAKE said:

Not sure which is harder, 42 perfect makuuchi placements or 63 perfect game winners in March Madness Tourney Pick’em

A perfect March madness bracket is probably harder in my mind but it's also a completely different kind of prediction.  Guessing the Banzuke is guessing what a small set of people decide in compromises with each other, which is quite different than predicting the results of 60+ basketball games (I don't know how many, but aren't there play-in games so there's more than 63?) involving hundreds of people who are competing. 

It's also impossible to guess the probabilities of either except in the absolute worst case scenario where you know nothing about the relative likelihoods of each outcome.  In such a situation, the Banzuke is actually much harder to guess (42! ~= 10^51, 2^63 ~= 10^19), but there is quite a lot that is absolutely impossible in Banzuke making, while nothing is impossible for basketball games.

Edited August 12, 2022 by Gurowake

[Sumo Spiffy 207]

1 hour ago, Gurowake said:

A perfect March madness bracket is probably harder in my mind but it's also a completely different kind of prediction.  Guessing the Banzuke is guessing what a small set of people decide in compromises with each other, which is quite different than predicting the results of 60+ basketball games (I don't know how many, but aren't there play-in games so there's more than 63?) involving hundreds of people who are competing. 

Play-in games are generally not included in bracket guesses, at least not the ones I've seen/used. Websites where you do bracket predictions generally give you until close to when the actual first round starts, so you have time to see who won the play-in games and adjust your bracket accordingly.

1 hour ago, Gurowake said:

It's also impossible to guess the probabilities of either except in the absolute worst case scenario where you know nothing about the relative likelihoods of each outcome.  In such a situation, the Banzuke is actually much harder to guess (42! ~= 10^51, 2^63 ~= 10^19), but there is quite a lot that is absolutely impossible in Banzuke making, while nothing is impossible for basketball games.

I was going to say something similar. Now that 16s occasionally beat 1s in the NCAA tournament, it really is 63 games with no essentially 100% outcomes, whereas some banzukes can have a single-digit number of slots that are actually questionable.

[Asashosakari 16,827]

17 hours ago, Sumo Spiffy said:

(...) whereas some banzukes can have a single-digit number of slots that are actually questionable.

Well, that's what people like to tell themselves, and then they score 43 points because actual committee practice is making a lot more slots uncertain...

[Sumo Spiffy 207]

14 hours ago, Asashosakari said:

Well, that's what people like to tell themselves, and then they score 43 points because actual committee practice is making a lot more slots uncertain...

Hey, I didn't say it happened a lot. >_>

[Kintamayama 40,590]

A week to go-the site is somewhat unstable again- do not wait any longer!! Mitakeumi stays at Ozeki! That should give you at least 2 points..

[Rocks 1,724]

I'm in, for a change. Although I expect this will be my worst score ever.

[Suwihuto 100]

Well that was a mess. There were a couple of rikishi that started about five ranks lower than I eventually placed them, with very little in the way of cogent ideas about why one or the other.

[Wamahada 24]

It is closed now - I wonder what the result will be