Test Run: Makushita Game

[Asashosakari 19,627]

The thing is, if "obvious picks" are those that ended up being picked a lot, I'm not so sure they're actually obviously good, and not just a result of the herd effect. For instance, I'm completely flabbergasted that Shibuya was picked the most in his bracket, and I'm ready to chalk that up to mere name recognition and not anything in his actual performances to date. Dewaotori seems a little "too obvious" to me, too, but we'll see I guess... I'll personally be happy to go 5-10. (Clapping wildly...)

It is an interesting question if successful sumo gaming means going with average picks. In other words, do players who pick rikishi that many other players pick are better than players with unusual picks (at least in games that do not rely on single bouts like ISP)?

I would say that very average picks generally lead to an average result, as well (neither very good nor very bad). Simply because lots of other player's picks diverge from yours only minimally, and by the very nature of (most) games, such a big cluster of players pretty much has to be found in the middle of the standings. To break out from the pack, your few non-average picks would all need to be really good ones, and that's obviously difficult to do.

Edited July 10, 2005 by Asashosakari

[sekihiryu 51]

Gokenzan looked to be in very poor shape

Gokenzan really should hang his mawashi somewhere. There is an imminent aura of sadness hovering in the air when he takes a decade to get back to bow after almost every other loss. Those knees must be completely wrecked and painful as hell. Long career already and doesn't look fit to get back to sekitori salary. He looks the most knee crippled of all rikishi if you take an average limping and general discomfort basho after basho. 1-6.

The popular selection that surprised me the most was Gokenzan, these 2 quotes are from our own Kaikitsune.

The man sounds very washed up, are people hoping for a miracle?

Danishiki's database was invalubale for this game, I plugged a lot of names into it

[Itachi 396]

Gokenzan looked to be in very poor shape

Gokenzan really should hang his mawashi somewhere. There is an imminent aura of sadness hovering in the air when he takes a decade to get back to bow after almost every other loss. Those knees must be completely wrecked and painful as hell. Long career already and doesn't look fit to get back to sekitori salary. He looks the most knee crippled of all rikishi if you take an average limping and general discomfort basho after basho. 1-6.

The popular selection that surprised me the most was Gokenzan, these 2 quotes are from our own Kaikitsune.

The man sounds very washed up, are people hoping for a miracle?

Danishiki's database was invalubale for this game, I plugged a lot of names into it

<{POST_SNAPBACK}>

I have to agree with this comment. I was really surprised that so many favoured this broken relic in that bracket. Nakao is pretty heavy too if you were going for size and sekitori experience but those guys are kind of old and were never very good in Juryo so why expect them back?

Hananosato, Hakuba, and Hoshikaze may be small but they are relatively young and may not have found their ceiling yet. Even Matsumidori seemed a better prospect to me. I have to admit I've given up on Hakuba but I may yet be proven wrong.

I don't see their matches though, so I could be way off.

[Itachi 396]

The thing is, if "obvious picks" are those that ended up being picked a lot, I'm not so sure they're actually obviously good, and not just a result of the herd effect. For instance, I'm completely flabbergasted that Shibuya was picked the most in his bracket, and I'm ready to chalk that up to mere name recognition and not anything in his actual performances to date. Dewaotori seems a little "too obvious" to me, too, but we'll see I guess... I'll personally be happy to go 5-10. (Nodding yes...)

Agreed, not obviously good in general - but in these game where most players are pretty far removed from the actual events, we mostly rely on the same sources - available to all online. These popular picks often reflect the 'media buzz' about the rikishi. Anyone who is really paying attention can learn that, for example, Sawaii is expected to do well even before he has his first bout in Jonokuchi.

When you see many of the top players agreeing on many of the picks it is because given the best information available to them, these are the obvious most likely best picks. We may vary from them a bit according to our hunches but to do consistently well we (those of us not in Japan and not able to watch the bouts regularly) should know who the best picks are according to the information freely available to everyone. I like to find a handful that have good buzz and then decide between them based on other factors that I come up with (trade secrets...)

Even in a game like Benchsumo, I find that if I select too many rikishi who are low in the selection stats, then I've made a mistake. This doesn't mean that all the popular picks are the best ones but very few of the unpopular rikishi have much success at all.

Speaking of stats, is there a site for this such as Oshirokita's GTYW site?

Could there be a compilation of selection stats to see how many picked each rikishi?

Edited July 11, 2005 by Itachi

[Randomitsuki 2,843]

Here are the standings after one day of action.

This stuff is counted manually, so errors are likely to occur.

I have received a 24th pick that was obviously sent before the deadline, but entered my mailbox after it.

As to the table: third column gives the score (i.e. surefire bracket winners or losers). The fourth column gives a trend, showing you how many of your rikishi are still on course for a win. In the fifth column AS stands for Alternative Scoring (or Asashosakari's Suggestion, if you will). By this statitics it is enough to finish among the top two of a bracket.

The suggested yusho tie-breakers apply for positioning (in case of a tie the higher ranked brackets are decisive).

It is surprising how many of us (including yours truly) are already firmly headed for MK-land after one seventh of the basho. The current trends are clear overestimates of the final score, and though your trend will sometimes move up a little bit, in general there will be a steady decline for each of us.

Makushita Game: Standings Nagoya 2005
Position	Shikona	Score	Trend	AS
1.	Susanoo	0-0	11-4	11-4
2.	Doitsuyama	0-0	11-4	11-4
3.	Torideyama	0-0	11-4	11-4
4.	Zenjimoto	0-0	11-4	11-4
5.	Itachi	0-0	11-4	11-4
6.	Flohru	0-0	10-5	10-5
7.	Kaihayaiha	0-0	9-6	9-6
8.	Kintamayama	0-0	9-6	9-6
9.	sekihiryu	0-0	9-6	9-6
10.	Sakana	0-0	9-6	9-6
11.	sashimaru	0-0	9-6	9-6
12.	Senpai	0-0	8-7	8-7
13.	Gusoyama	0-0	8-7	8-7
14.	Takamueda	0-0	8-7	8-7
15.	Oshirokita	0-0	7-8	7-8
16.	Gernobono	0-0	7-8	7-8
17.	Asashosakari	0-0	7-8	7-8
18.	Randomitsuki	0-0	7-8	7-8
19.	Feginowaka	0-0	6-9	6-9
20.	Hamanoyama	0-0	5-10	5-10
21.	Yubiquitoyama	0-0	5-10	5-10
22.	Holleshoryu	0-0	5-10	5-10
23.	Tamanaogijima	0-0	4-11	4-11
24.	Hakanonami	0-0	4-11	4-11

Edited July 11, 2005 by Randomitsuki

[Torideyama 32]

Wow,

I am in first place (tied) on the first day of a first time sumo game. I guess I should print a copy of the results and frame them, cause it ain't gonna be this way for long.

Torideyama

[Randomitsuki 2,843]

The standings after two days.

And as you can see, we are having some problems already. According to the original scoring method each and every one of is on firm course for a make-koshi with five score-diminishing days still to come. Even the alternative method of scoring doesn't make us look much better.

I know it is too early, but we might need another scoring method. Here are three solutions for discussion (in almost no particular order):

1) Getting into the top 3 of each bracket instead of the top 2. But this would really make us look like amateurs.

2) Having 20 brackets instead of 15, and naming 20 rikishi for a basho.

3) Following Doitsuyama's suggestions and just look at the wins (like RotoSumo). However, then we wouldn't need 15 brackets to begin with.

Makushita Game: Standings Nagoya 2005
Position	Shikona	Score	Trend	AS
1.	sekihiryu	0-0	7-8	8-7
2.	Doitsuyama	0-0	7-8	7-8
3.	Zenjimoto	0-0	7-8	7-8
4.	Susanoo	0-0	6-9	7-8
4.	sashimaru	0-0	6-9	7-8
6.	Flohru	0-0	6-9	7-8
7.	Yubiquitoyama	0-0	5-10	5-10
8.	Asashosakari	0-0	5-10	6-9
9.	Randomitsuki	0-0	5-10	5-10
10.	Kaihayaiha	0-0	5-10	6-9
11.	Torideyama	0-0	4-11	5-10
12.	Senpai	0-0	4-11	5-10
13.	Itachi	0-0	4-11	5-10
14.	Oshirokita	0-0	4-11	4-11
15.	Kintamayama	0-0	4-11	5-10
16.	Hamanoyama	0-0	4-11	5-10
17.	Feginowaka	0-0	4-11	4-11
18.	Gusoyama	0-0	3-12	4-11
19.	Holleshoryu	0-0	3-12	3-12
20.	Sakana	0-0	3-12	4-11
21.	Takamueda	0-0	3-12	4-11
22.	Tamanaogijima	0-0	2-13	2-13
23.	Hakanonami	0-0	1-14	2-13
24.	Gernobono	0-0	1-14	2-13

[Oshirokita 192]

Well, knowing how hard it is to pick yusho winners (or even just a kachi koshi rikishi) from an entire division, I suspect it is quite difficult to get the best in 15 different sub-brackets. I definitely like the idea of choosing from 15 brackets as it really makes you think about every Makushita rikishi. It might be an idea to score the game by giving a win for every kachi koshi you successfully pick. This is still unlikely to give a 15-0. Then you could tally how many "best picks" you got and possibly use that number as a tie-breaker. But we'll see just how badly we do under the current system test.

[Asashosakari 19,627]

Are you sure your trend overestimates the eventual results? Given that the top score in most brackets will be 5-2, I don't think having your picks at 1-1 now instead of 2-0 is that much of a sign of impending doom. Nevermind, I see your point now. I really shouldn't try to juggle probabilities in my head while half-asleep. (Nodding yes...) I do think the alternative "within two" scoring won't be quite as susceptible to diminishing scores during the course of the basho, though. (I just hope we players manage to outscore random chance...)

It might be an idea to score the game by giving a win for every kachi koshi you successfully pick.

I think that would be too easy. You should be able to pick half KK rikishi just by chance, after all.

Edited July 13, 2005 by Asashosakari

[Doitsuyama 1,192]

I still think just counting the wins is fair and easy. No difficulties with scoring here. (If you want to avoid the effect of a 1-6 or 0-0-7 from a bad pick, we could even allow for one mulligan, counting only the 14 best rikishi in a selection. Not really necessary though.) I like the idea of 15 brackets. Just selecting any 15 Makushita rikishi is not a bad idea though.

[Sakana 46]

day one : 9-6 -> day two : 3-12... OUCH ! O_o!

[Zenjimoto 40]

The 15 brackets are what makes this game tick in the first place. I would suggest renaming the game Toriteki-Toto and getting a win for each bracket where your pick is above the median line ;-)

But maybe we really ought to see how the scoring works out, first? (Applauding...)

Just joking on the name... (Applauding...)

Cheers

Zenjimoto

[Randomitsuki 2,843]

Standings after three bouts per Makushita rikishi

With all brackets except for two having just a single 3-0 rikishi in the lead, the trend column currently underestimates the final results. As a consequence, almost every 2-1 rikishi is currently tied for second place in a bracket, therefore the Alternative Scoring (AS) column currently is an overestimation.

Makushita Game: Standings Nagoya 2005
Position	Shikona	Score	Trend	AS
1.	Hamanoyama	0-0	4-11	7-8
2.	Kaihayaiha	0-0	4-11	9-6
3.	Torideyama	0-0	3-12	9-6
4.	Yubiquitoyama	0-0	3-12	6-9
5.	Asashosakari	0-0	3-12	9-6
6.	Susanoo	0-0	3-12	10-5
7.	Itachi	0-0	3-12	8-7
8.	Zenjimoto	0-0	3-12	9-6
9.	sekihiryu	0-0	3-12	9-6
10.	sashimaru	0-0	3-12	9-6
11.	Doitsuyama	0-0	2-13	10-5
12.	Oshirokita	0-0	2-13	5-10
13.	Takamueda	0-0	2-13	6-9
14.	Feginowaka	0-0	2-13	6-9
14.	Gusoyama	0-0	2-13	7-8
14.	Randomitsuki	0-0	2-13	7-8
14.	Sakana	0-0	2-13	8-7
18.	Kintamayama	0-0	1-14	7-8
18.	Senpai	0-0	1-14	5-10
20.	Flohru	0-0	1-14	8-7
21.	Gernobono	0-0	1-14	4-11
21.	Tamanaogijima	0-0	1-14	4-11
23.	Holleshoryu	0-0	1-14	7-8
24.	Hakanonami	0-0	0-15	4-11

[Randomitsuki 2,843]

Standings after four bouts per Makushita rikishi

Makushita Game: Standings Nagoya 2005
Position	Shikona	Score	Trend	AS
1.	Hamanoyama	0-0	6-9	7-8
2.	Torideyama	0-0	5-10	8-7
3.	sekihiryu	0-0	5-10	7-8
4.	Randomitsuki	0-0	5-10	5-10
5.	Kaihayaiha	0-0	5-10	6-9
6.	Yubiquitoyama	0-0	4-11	6-9
7.	Asashosakari	0-0	4-11	7-8
7.	Doitsuyama	0-0	4-11	8-7
9.	Susanoo	0-0	4-11	9-6
10.	Gusoyama	0-0	4-11	4-11
11.	Kintamayama	0-0	4-11	5-10
12.	Zenjimoto	0-0	4-11	7-8
13.	Oshirokita	0-0	3-12	7-8
14.	Sakana	0-0	3-12	3-12
15.	Feginowaka	0-0	3-12	4-11
16.	Flohru	0-0	3-12	6-9
17.	Gernobono	0-0	3-12	3-12
18.	Takamueda	0-0	2-13	4-11
19.	Senpai	0-0	1-14	6-9
20.	sashimaru	0-1	5-10	8-7
21.	Itachi	0-1	3-12	5-10
22.	Tamanaogijima	0-1	3-12	4-11
23.	Holleshoryu	0-1	2-13	4-11
24.	Hakanonami	0-1	1-14	4-11

[Takamueda 0]

Banzuke making in this game could be VERY difficult if the results don't get better...

Edited July 17, 2005 by Takamueda

[Itachi 396]

Banzuke making in this game could be VERY difficult if the results don't get better...

<{POST_SNAPBACK}>

Perhaps we could all start as Tsukedashi and drop down from there...

[Randomitsuki 2,843]

Every Makushita rikishi had five bout until now.

Once more there is a new option in the standings. The three columns now indicate how your standings are if only the best rikishi in a bracket was picked (the original idea, now named "Top 1"), if your rikishi was among the two best in a bracket (Top 2) or in the top 3.

Since the final scoring method is not determined yet, I have ranked players according to the sum of the three scores.

Makushita Game: Standings Nagoya 2005
Position	Shikona	Top 1	Top 2	Top 3
1.	Susanoo	5-10	9-6	10-5
2.	Torideyama	4-11	7-8	12-3
3.	Hamanoyama	5-10	8-7	9-6
4.	Doitsuyama	5-10	7-8	9-6
5.	Oshirokita	4-11	7-8	9-6
5.	sashimaru	4-11	7-8	9-6
7.	sekihiryu	4-11	6-9	10-5
8.	Asashosakari	3-12	7-8	9-6
9.	Yubiquitoyama	3-12	7-8	8-7
10.	Kaihayaiha	4-11	6-9	7-8
10.	Randomitsuki	4-11	6-9	7-8
12.	Senpai	2-13	7-8	8-7
13.	Zenjimoto	4-11	5-10	7-8
14.	Flohru	3-12	6-9	7-8
15.	Itachi	4-11	5-10	6-9
16.	Takamueda	3-12	5-10	7-8
17.	Tamanaogijima	4-11	4-11	6-9
18.	Feginowaka	3-12	5-10	6-9
19.	Sakana	3-12	4-11	6-9
20.	Kintamayama	2-13	4-11	6-9
21.	Hakanonami	1-14	5-10	6-9
22.	Gusoyama	2-13	4-11	5-10
23.	Gernobono	1-14	2-13	4-11
24.	Holleshoryu	0-15	3-12	4-11

[Doitsuyama 1,192]

If you for the top 3 method in the end (which doesn't look bad so far), a special prize could be given for the player with the most top 1 rikishi.

[Randomitsuki 2,843]

After six rounds of action, current scoring is quite high for all scoring methods. This is partially due to bracket 7 where currently five rikishi are in the lead with 3-3!

Three more days, and we know so much more.

Makushita Game: Standings Nagoya 2005
Position	Shikona	Top 1	Top 2	Top 3
1.	Susanoo	5-10	10-5	11-4
2.	Doitsuyama	6-9	9-6	10-5
3.	sekihiryu	5-10	9-6	11-4
4.	Torideyama	4-11	10-5	11-4
5.	sashimaru	6-9	8-7	8-7
5.	Itachi	5-10	8-7	9-6
7.	Kaihayaiha	6-9	7-8	8-7
8.	Asashosakari	5-10	6-9	10-5
9.	Hamanoyama	5-10	7-8	8-7
10.	Senpai	3-12	8-7	9-6
11.	Feginowaka	5-10	6-9	8-7
11.	Yubiquitoyama	5-10	6-9	8-7
13.	Zenjimoto	4-11	7-8	8-7
14.	Randomitsuki	4-11	6-9	9-6
15.	Gusoyama	4-11	6-9	8-7
16.	Oshirokita	3-12	7-8	8-7
17.	Flohru	5-10	6-9	6-9
17.	Takamueda	5-10	6-9	6-9
19.	Sakana	4-11	5-10	8-7
20.	Kintamayama	3-12	6-9	7-8
21.	Tamanaogijima	2-13	5-10	8-7
22.	Hakanonami	1-14	5-10	5-10
23.	Holleshoryu	2-13	4-11	4-11
24.	Gernobono	1-14	2-13	4-11

Edited July 21, 2005 by Randomitsuki

[Itachi 396]

Wow! I went from 15th to 5th between the last two updates! Just when I was starting to hang my head in shame - I can't very well go mk in the game I begged for!

[Randomitsuki 2,843]

Well then, here are the preliminary final results of the Makushita Game in Nagoya. According to the original scoring method the best record would have been by Yubiquitoyama (6-9). Taking the top two spots into account we still would have too many MK (22 vs. 2 KK). So until further notice I take the standings from the top 3 to be the final standings of this basho. That gives you 12 KK and 12 MK.

In case of a tie the higher bracket placements were the decider.

All this was done manually, so there might be some errors. I will double-check results when I have the time for it.

For sake of completeness I have included the alternative scoring methods (Top 2; Top 1)

Makushita Game: Standings Nagoya 2005
Position	Shikona	Top 3	Top 2	Top 1
1.	Susanoo	11-4	9-6	5-10
2.	Doitsuyama	11-4	7-8	5-10
3.	sekihiryu	11-4	6-9	1-14
4.	Itachi	10-5	7-8	4-11
5.	Hamanoyama	9-6	8-7	4-11
6.	Oshirokita	9-6	4-11	3-12
7.	Torideyama	9-6	6-9	3-12
8.	Zenjimoto	9-6	7-8	3-12
9.	Kaihayaiha	9-6	6-9	5-10
10.	Yubiquitoyama	8-7	7-8	6-9
11.	Randomitsuki	8-7	4-11	2-13
12.	sashimaru	8-7	6-9	5-10
13.	Senpai	7-8	5-10	2-13
14.	Tamanaogijima	7-8	5-10	0-15
15.	Asashosakari	7-8	5-10	4-11
16.	Kintamayama	7-8	5-10	3-12
17.	Sakana	7-8	4-11	2-13
18.	Gusoyama	7-8	5-10	3-12
19.	Flohru	7-8	4-11	3-12
20.	Feginowaka	7-8	5-10	3-12
21.	Takamueda	6-9	4-11	4-11
22.	Hakanonami	6-9	4-11	2-13
23.	Holleshoryu	6-9	3-12	3-12
24.	Gernobono	5-10	3-12	1-14

[Yubiquitoyama 4]

I think it's obvious it's the "top-1"-method that somehow should be used ;-)

[Asashosakari 19,627]

I think it's obvious it's the "top-1"-method that somehow should be used ;-)

<{POST_SNAPBACK}>

Because nothing says "success" like having the best losing record? ;-)

[Yubiquitoyama 4]

I think it's obvious it's the "top-1"-method that somehow should be used ;-)

<{POST_SNAPBACK}>

Because nothing says "success" like having the best losing record? ;-)

<{POST_SNAPBACK}>

No, but you see if I had known that it was the best 3 that would be counted, my entry would have looked COMPLETELY different ;-) ;-)

[Itachi 396]

My suggestion is that we take the sum of our top 1, top 2, and top3 records.

This would give Yubi-zeki 3 points for each of his game best 6 correct picls and reduce the punishment he takes for only finding 2 other top 3 finishers.

I believe Susanoo would still be the winner but those who failed to find many Top 1 picks would not be disproportionately rewarded.

With only 8 rikishi per bracket, finding the top 3 is only one away from a 50/50 guess. No wonder the mk/kk line is right at 50/50 when done this way.

What would the standings look like with the sum method?