Kyokutenho to set another "record"

[Gurowake 4,109]

Assuming the DB is accurate, Kyokutenho tied Terao for the most losses in Ozumo last basho. Check it out: http://sumodb.sumogames.de/Query_bout.aspx?show_form=0&group_by=rikishi1&onlyl1=on It seems reasonable to assume that he'll set it this basho as he's already on the torikumi and thus would require a zensho to avoid setting it. He's been the leader in total bouts for a few basho, being 41 ahead of Terao now.

He's still fairly far away from Terao's record of most losses as Makuuchi though, 753 to 714, and is also behind Takamiyama in that regard. He's almost 100 off the record for Makuuchi bouts, sitting in 4th behind Kaio and the other two large-loss rikishi mentioned previously.

Sorry if this has been mentioned elsewhere, but Google didn't find anything about it despite picking up other posts from here when looking for "Kyokutenho record loss." The forum search function is totally useless.

I did the original search that found the first-mentioned tie after reading the Persistence Watch thread and wondering who had the most losses, not realizing it probably wasn't someone who had only half the fights per basho of sekitori until the results came up.

[Asashosakari 19,669]

You may want to take a look at the full career profiles for Terao and Kyokutenho...

[Gurowake 4,109]

Well, the opening clause of my post conditioned the entire analysis on the database being accurate, and indeed the bout database is not nearly as complete as I assumed it was. Now that I've looked through Terao's record, I see that there are basho where his record has been entered and it is known on which days he won and lost, but there's no knowledge of who he fought and thus the bout is not in the bout database. The database is a great resource that I forget the limitations of; I just had assumed all the data was there from any time period during my lifetime given how I could see which days they won and lost (which I do know stops sometime like in the 60s), but now it looks like they just remembered that information and not who they fought.

It's possible there's enough information for basho like Kyushu 1979 in the days each rikishi won/lost to reconstruct the actual torikumi for the lower divisions using knowledge of how it would have been constructed, being able to distinguish from different possibilities based on the win/loss pattern of successive days. If I get really bored some day I might try it, but it definitely would be better suited to computer analysis due to the sheer amount of data. I could probably write pseudo-code given enough effort and desire, but I know absolutely nothing about actual coding. I understand that even if one could make out a consistent set of torikumi for all days that looked reasonable there's no guarantee that it's the one that they actually used; it would be an interesting project regardless.

Edited July 13, 2014 by Gurowake

[Vikanohara 171]

Speaking about records, I wonder which record Hakuho is set to break this basho. :-P

On a more serious note, it would be interesting to have one topic focussing on all kinds of records.

[Doitsuyama 1,192]

It's possible there's enough information for basho like Kyushu 1979 in the days each rikishi won/lost to reconstruct the actual torikumi for the lower divisions using knowledge of how it would have been constructed, being able to distinguish from different possibilities based on the win/loss pattern of successive days. If I get really bored some day I might try it, but it definitely would be better suited to computer analysis due to the sheer amount of data. I could probably write pseudo-code given enough effort and desire, but I know absolutely nothing about actual coding. I understand that even if one could make out a consistent set of torikumi for all days that looked reasonable there's no guarantee that it's the one that they actually used; it would be an interesting project regardless.

This is actually how I entered all the lower division torikumi. The lower division torikumi indeed are highly predictable, but not 100% especially because same-stable or already-fought road blocks are dealt with in different ways. It helps a lot that the bouts are split in two days though. Deviations on each day magnify subsequent errors obviously, so I ran the algorithm for each pair of days, checked with the available torikumi, ran the next day and so on. Still a lot of work, but much easier than typing ALL bouts.

I thought about running the algorithm without check for those basho where only the hoshitori-hyo is available but no actual pairings but quickly dropped it because there simply would be too many errors.

[Gurowake 4,109]

Yes, thinking more about it, designing an algorithm that could fluidly make reasonable guesses as to potential ways the torikumi could go, and then verify that it's consistent across the whole thing is somewhat dependent on the situations that need addressing being isolated enough from each other; while the torikumi for the lower divisions for Days 1 and 2 have been consistent at breaking the pairing of the match lower in rank of one that would be initially scheduled between heya-mates and pairing 1v3 and 2v4, I can see once you get to considering the 3rd and successive matches there things might get quite complicated.

And of course sometimes they just do weird stuff. Day 2 of the current basho in Juryo, there would have been 9e v 10w had they not been in the same heya, causing the bottom part of the torikumi to no longer be as regular as the top part. Looking at it, there's a duplicated pattern in the last 6 matches: Ne v N+2e, Nw v N+1w and N+1w v N+2w for N = 9 and 12, but wouldn't Ne v N+1e, Nw v. N+2e, N+1w v. N+2w make more sense in terms of keeping people against those closer ranked to them? Differences in rank are (2, 1, 1) vs. (1, 1.5, 1). I guess I haven't looked as records, and since (I think) they do the torikumi after the Juryo matches they might take the results into consideration and go with the alternative that matches people up best, but given how regular and disrespective of record the rest of the schedule is it seems unlikely.

So yeah, probably not a great idea to look into further; but it certainly has made me more interested in how they do the lower division torikumi, as simple as it normally is.

[yorikiried by fate 2,029]

There was once a thread touching such a topic.