Sign in to follow this  
Guest Vincentohana

ODD SUMO

Recommended Posts

Guest Vincentohana

Hello Everybody,

Do you remember oddsumo game ? I tried to contact the webmaster Kojamuri but I didn't have any answer. Do you know how can I do to join Kojamuri ? I want to take again oddsumo myself

Thanks for you answer

Sorry for my english

Share this post


Link to post
Share on other sites

I'm bumping this up- is anyone planning on finally renewing this game? I remember Gusoyama talking about it.

Come on people, let's get it ready for Aki-it can't be THAT complicated- there are geniuses out there who have done more in less..

Share this post


Link to post
Share on other sites

I've poked around with it, but it definitely won't be ready for Aki. My grabber's broken right now. I'll take a look after I do the Aki banzuke for SCS.

By all means, if someone else wants to ressurect this game, then go for it! I hold no patent or claim on this, it was just one of my most favorite games, and Im sad to see it go.

Share this post


Link to post
Share on other sites

It''s annoying that the entire framework of the game is basically up and running, and should be able to be hijacked with just some minor changes to the grabber (or by all means, why not enter those few results manually, even!).

I tried to reach Kojamuri (Wieslaw Schupke) several times myself, to no avail.... perhaps some of the people in the german speaking forum (I am looking at Kana-san in particular) have some ideas about this. I miss this game dearly, as well. It was a brilliant concept and a very loveable whimsical implementation. Would be great to see it back!

Cheers

Z.

Share this post


Link to post
Share on other sites

Currently I have many things to do for Sekitori Oracle and Quad, but if noone starts working on resurrecting Odd Sumo within next 2-3 months then I think I could take care of it.

But to do it I would need precise info about all the rules. Especially how to calculate these odds because I have no idea how it works.

Share this post


Link to post
Share on other sites

Count my vote for resurecting Oddsumo aswell......

Share this post


Link to post
Share on other sites

My strategy for Oddsumo was a total failure. I only won a few bets in my first basho and I think nothing after that.

Still, I would be interested in giving it another go with a fresh approach.

I'd also like to know how the odds were arrived at.

Share this post


Link to post
Share on other sites
But to do it I would need precise info about all the rules. Especially how to calculate these odds because I have no idea how it works.

the rules say:

he single quotas are calculated by

- the differences between ranks of the banzuke

- the current basho results of each rikishi

- the last contest between the two rikishi

and the quota are known for every day of the last basho.....

i guess there has to be some statistically/mathematically trained member of the forum to find the secret formula through reverse engineering...

Share this post


Link to post
Share on other sites

Thanks. This info looks promising.

I am sure there are many members on the forum who could provide a formula.

The question is who will provide the best one (I am not worthy...)

Share this post


Link to post
Share on other sites
the rules say:

he single quotas are calculated by

- the differences between ranks of the banzuke

- the current basho results of each rikishi

- the last contest between the two rikishi

and the quota are known for every day of the last basho.....

i guess there has to be some statistically/mathematically trained member of the forum to find the secret formula through reverse engineering...

I gave it a quick try, but either the rank difference component is dynamic, or there's something else that influences things:

Day 4: M4w Takamisakari (1-2) - M2w Kakizoe (1-2), first-ever meeting, odds 1.50 for both

Day 5: M7e Kokkai (3-1) - M9e Kotoryu (3-1), first-ever meeting, odds 1.45 for Kokkai and 1.55 for Kotoryu

Same records, same rank difference, no prior matchup, but still different odds.

Share this post


Link to post
Share on other sites

I also fiddled around with the stats on the site, but couldn't make sense of it. Initially I thought that the previous encounter increases the odds of the last winner by 0.05 (consider Takamisakari vs. Tokitsuumi on day 1 - both M4, both with 0-0 record, and Sakari with a 1.45 vs. 1.55 because he won the last bout between these two). But there was the similar case of Aminishiki vs. forgot-whom with very similar situation (Aminishiki won the last one, both were M11 with a 0-0 record), but this resulted in a 1.5 vs. 1.5 tie for the odds.

Actually I'd suggest developing a new algorithm for making odds. Seems to be much easier.

Share this post


Link to post
Share on other sites

A was playing this game for 1 or 2 bashos and I got feeling that odds were calculated just "by eye" and not by some formula

Share this post


Link to post
Share on other sites

I think that it would be tricky to make a reliable formula based on stats alone.

For example, on day 1 of a basho, Hakuho meets Plobadob ranked at K1W. Plobadob got promoted to that rank (with a bit of banzuke luck) after a 10-5 record at M5. He has never previously faced the Yokozuna.

Therefore head-to-heads are equal, current basho records are equal, and they are only two ranks apart. Using stats alone might make the odds closer to 'even' than an informed eye alone would have them.

Share this post


Link to post
Share on other sites
I think that it would be tricky to make a reliable formula based on stats alone.

For example, on day 1 of a basho, Hakuho meets Plobadob ranked at K1W. Plobadob got promoted to that rank (with a bit of banzuke luck) after a 10-5 record at M5. He has never previously faced the Yokozuna.

Therefore head-to-heads are equal, current basho records are equal, and they are only two ranks apart. Using stats alone might make the odds closer to 'even' than an informed eye alone would have them.

Well, firstly komusubi and yokozuna are three ranks apart, not two (like 2009 is one year from now, not two :-D ). Next, the rank difference component can be dependent on the ranks, not just the number difference. Which means a matrix of deductions with the two ranks on the two axis of the matrix, so Y-K deduction can be quite different from M5-M1 deduction for example. I suspect the ranks are grouped which would also explain Asashosakari's finding of different deductions for same rank difference (explained with same rank group in the first case, different rank groups in the second case).

Edited by Doitsuyama

Share this post


Link to post
Share on other sites
Next, the rank difference component can be dependent on the ranks, not just the number difference. Which means a matrix of deductions with the two ranks on the two axis of the matrix, so Y-K deduction can be quite different from M5-M1 deduction for example. I suspect the ranks are grouped which would also explain Asashosakari's finding of different deductions for same rank difference (explained with same rank group in the first case, different rank groups in the second case).

That's quite possible...there was also an O2e-M2e (or something like that) bout in my quickly surveyed sample, and the (equivalent) rank difference going into the calculation seemed to be more like 9 or 10, rather than 4, so there's probably some accelerated structure to the deductions indeed.

Share this post


Link to post
Share on other sites
I think that it would be tricky to make a reliable formula based on stats alone.

For example, on day 1 of a basho, Hakuho meets Plobadob ranked at K1W. Plobadob got promoted to that rank (with a bit of banzuke luck) after a 10-5 record at M5. He has never previously faced the Yokozuna.

Therefore head-to-heads are equal, current basho records are equal, and they are only two ranks apart. Using stats alone might make the odds closer to 'even' than an informed eye alone would have them.

Well, firstly komusubi and yokozuna are three ranks apart, not two (like 2009 is one year from now, not two :-D ). Next, the rank difference component can be dependent on the ranks, not just the number difference. Which means a matrix of deductions with the two ranks on the two axis of the matrix, so Y-K deduction can be quite different from M5-M1 deduction for example. I suspect the ranks are grouped which would also explain Asashosakari's finding of different deductions for same rank difference (explained with same rank group in the first case, different rank groups in the second case).

Agreed, but this is not my point.

If you focused on rank alone, you could have the situation where rikishi X is ranked at M15 facing rikishi Y who is at M10. However the head-to-heads are currently 10-2 in favour of rikishi X. And rikishi X is currently 5-1 in the basho, whereas rikishi Y is at 2-4. I would say that in this instance rank is probably not the best indicator of the likely outcome of the bout. Therefore past head-to-heads and basho results are also important to consider.

But if (for example) Ama were to be ranked at K1W (maybe he goes 7-8 at the next basho), and were to face Hakuho on shonichi, if we were to go by the rank difference only, he would be given the same odds of defeating the Yokozuna as Plobadob (if ranked at the same spot). I would think that the 'informed eye' would take on board Ama's past experience and shorten the odds for him against the Y compared to Plobadob.

Now, if you factored in past head-to-heads and basho results to date the odds for Plobadob would be shorter, because although they have the same basho record (0-0) going into the bout, and the same rank difference, Ama has a worse head-to-head record against the Yokozuna.

Therefore a formula based on the three factors - or on just one or two of them (or with heavy weight on one aspect) would throw up some 'strange' odds that might need tinkering with by the 'informed eye' from time-to-time.

I have an alternative suggestion. Perhaps the formula for making the odds could be based on Doitsuyama's Ozumo Power Ratings?

Edit: Tidied up my grammar a little.

Edited by Jejima

Share this post


Link to post
Share on other sites

I think using the 'informed eye' is supposed to be the element for mastering the game. Using my ratings (maybe with some room to spare for the benefit of the bank but not really necessary) would make it quite difficult to separate good from bad players - because it basically wouldn't matter anymore which side you are betting. Using my ratings would be good from the view of the bank, to maximize the profit - but not good to develop a game to find good and bad players.

Edited by Doitsuyama

Share this post


Link to post
Share on other sites

I'd say it's also quite an unlikely scenario that a guy who got a lucky komusubi promotion just so happens to be a rikishi who has zero meatgrinder experience at all, especially if we're also assuming that he'd be a rikishi who doesn't deserve to be up there. I'd say it's much more likely that somebody who takes the banzuke by storm in that manner is exactly the type of rikishi who should be receiving near-even odds against high-ranking opponents he has never faced.

Quick edit: Oh, like that Asashoryu guy, for instance.

Edited by Asashosakari

Share this post


Link to post
Share on other sites
I think using the 'informed eye' is supposed to be the element for mastering the game. Using my ratings (maybe with some room to spare for the benefit of the bank but not really necessary) would make it quite difficult to separate good from bad players - because it basically wouldn't matter anymore which side you are betting. Using my ratings would be good from the view of the bank, to maximize the profit - but not good to develop a game to find good and bad players.

I respectfully disagree. If we say that using your ratings are the closest we can get to true odds, then we should use those. The 'bad players' will look no further. The 'good players' will take other things into consideration, and this will give them the edge.

If we use a formula that sometimes does not come close to reflecting the true odds, then a very wrongly weighted bout should be easy to spot for 'good' and 'bad' players alike, and everyone will jump on the same band-wagon.

Share this post


Link to post
Share on other sites
I respectfully disagree. If we say that using your ratings are the closest we can get to true odds, then we should use those. The 'bad players' will look no further. The 'good players' will take other things into consideration, and this will give them the edge.

I think Doitsuyama's point is that "the edge" to be achieved will be sufficiently small under these circumstances that it will be totally swamped by luck. I don't think a game where a truly good player might still only have a 55-60% chance of kachi-koshi would be very fun to play in the long run.

Share this post


Link to post
Share on other sites
I'd say it's also quite an unlikely scenario that a guy who got a lucky komusubi promotion just so happens to be a rikishi who has zero meatgrinder experience at all, especially if we're also assuming that he'd be a rikishi who doesn't deserve to be up there. I'd say it's much more likely that somebody who takes the banzuke by storm in that manner is exactly the type of rikishi who should be receiving near-even odds against high-ranking opponents he has never faced.

Quick edit: Oh, like that Asashoryu guy, for instance.

Yes, you are correct. It is an unlikely scenario.

This chap Ryogoku is almost a counter-example, except he defeated a few ozekis in the basho (8-7 at M5) prior to meeting the Yokozuna for the first time at K1W, so he did have 'some form'.

There may be one or two others, but I think you are right in saying that it is generally more likely to be an 'Asashoryu-type' than anything else, so closer odds are justified.

The question is would everyone be able to spot if it is an Asashoryu or a Ryogoku? If yes, then perhaps my argument still stands. If no, then Doitsuyama's suggestion of the correct usage of the 'informed eye' is probably more correct. (I am not worthy...)

But that is only one example......

When I played the new Sumo trading game in May, I was able to do quite well due to the fact that I was not really betting on what I thought would happen, but rather betting whenever I thought the odds were too skewed in one direction (even if I suspected the other outcome was more likely to happen). Thinking a bit more, this would suggest that perhaps other players do not always take these into consideration, so therefore mildly inaccurate odds would help players like me :-D

Edit: Put the link in correctly :-)

2nd Edit: 'Thinking a bit more....' etc

Edited by Jejima

Share this post


Link to post
Share on other sites
But to do it I would need precise info about all the rules. Especially how to calculate these odds because I have no idea how it works.

the rules say:

he single quotas are calculated by

- the differences between ranks of the banzuke

- the current basho results of each rikishi

- the last contest between the two rikishi

and the quota are known for every day of the last basho.....

i guess there has to be some statistically/mathematically trained member of the forum to find the secret formula through reverse engineering...

My understanding at the time was that Odd Sumo was picking up the odds from a Japanese betting site and not calculating its own odds. Not all bouts were available for betting IIRC. I thought that part of the demise of the game was loss of access to the Japanese odds.

Warm regards,

Kofuji

Share this post


Link to post
Share on other sites
I'd say it's also quite an unlikely scenario that a guy who got a lucky komusubi promotion just so happens to be a rikishi who has zero meatgrinder experience at all, especially if we're also assuming that he'd be a rikishi who doesn't deserve to be up there. I'd say it's much more likely that somebody who takes the banzuke by storm in that manner is exactly the type of rikishi who should be receiving near-even odds against high-ranking opponents he has never faced.

Quick edit: Oh, like that Asashoryu guy, for instance.

Yes, you are correct. It is an unlikely scenario.

Well, even if the unlikely scenario is happening I don't see the point why it should be bad for the game. If I rememeber correctly, every day in Odd Sumo there were a few bouts with odd odds :-D (e.g. a Yokozuna getting too good odds) but of course every player could cash in on that (or lose everything). Furthermore you had to bet on at least three different bouts a day which relativised one "easy" bout with good odds. And of course the most interesting point in the game always was if you make e.g. a three-bout-bet with a good chance to win or if you take the risk and try a ridiculous six-bout-bet that would win the basho for you at once (wasn't that Gernobonos strategy?!).

So I vote for keeping the odd-making process exactly the way it was (if anyone can remember for sure HOW it actually was) (I am not worthy...)

Edited by Flohru

Share this post


Link to post
Share on other sites

Here's how I get my odds:

East chances of winning: (.5*e rank+1.3*#of e wins + 2*head to head advantage for e)/((.5*e rank+1.3*#of e wins + 2*head to head advantage for e)(.5*w rank+1.3*#of w wins + 2*head to head advantage for w)

If that makes any sense. In short, it counts head to head heaviest, then record, then rank. I was thinking of adding in overall career record somewhere in there.

Here's the URL to my odds, you can substitute any year/month/day combo that you wish. Gusoyama Odd Sumo torikumi

Share this post


Link to post
Share on other sites

I take back everything I said, these odds look workable (although Asashoryu vs Toyonoshima on day 1 with 95.6% vs 4.4% does look a little lopsided.)

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