Rikishi Strength Analysis Revisited

[Doitsuyama 1,173]

Some of you might remember my Historical Strength Analysis from one year ago. If not it might be a good idea to go back to that topic, because I won't repeat the explanation of the methodical approach, table column meanings etc. since nothing of these have changed this time.Six basho have gone by since this last edition, and I thought it's a good time to look at the changes. Especially since some rikishi's performances, Hakuho and Miyabiyama on the forefront, are throwing up questions in historical context lately.The following tables are listing all rikishi with career high rank in makuuchi who were active in the last six basho, grouped by yokozuna, ozeki, sekiwake, komusubi and maegashira, just like the original tables. Since the rikishi who weren't sekitori after Nagoya 2005 can't change strength ratings I omitted all those for the sake of better readability. You can refer to the original analysis for these (their ranks can be lower now of course with new rikishi rising).Let's start right away with the simplest category as Asashoryu stayed the sole active yokozuna all year. I actually have added two new columns on the right which serve to illustrate the change over the last year. Δ P is the change of the Peak rating (always positive or 0) and Δ R10 is the change of the Rating ₁₀ (normally positive or 0, but can actually get negative as well for rikishi in decline).

Yokozuna 1960-2006

Rank	Shikona	First	Last	Basho	Makuuchi	Kyujo	W-L-D	Peak	Rank	Rating 3	Rank	Rating 6	Rank	Rating 10	Rank	Δ P	Δ R10
Y	Asashoryu	200009	200607	36	34	0	410-114	2612	5	2606	5	2582	5	2531	9	4	75

Asashoryu was able to improve his peak slightly after the 14-1 in Kyushu, but suffered an 11-4 in Hatsu and an 1-2-12 in Natsu. Each time he recovered quickly with yusho in Haru and Nagoya of course, but still is 30 short of his peak right now. The last year underlined his career consistency with those two basho being mere bumps on the road in the big picture and his R3, R6 and R10 values have risen considerably. With his peak rating he clipped ahead of Kitanofuji and his R3 and R6 also are ranked 5th just ahead of Kitanofuji, behind only Taiho, Kitanoumi, Tamanoumi and Chiyonofuji. These four are still considerably ahead of Asashoryu except Tamanoumi's R6, which would be better if his career hadn't been tragically cut short. Asahoryu's R10 is lower only because his career still is relatively short and will rise over the next year even if he has an off year.

Ozeki 1960-2006

Rank	Shikona	First	Last	Basho	Makuuchi	Kyujo	W-L-D	Peak	Rank	Rating 3	Rank	Rating 6	Rank	Rating 10	Rank	Δ P	Δ R10
O	Hakuho	200401	200607	16	14	0	168-65	2512	25	2442	41	2365	58	2317	69	168	106
O	Tochiazuma	199605	200607	57	54	5	480-275	2463	39	2450	36	2428	37	2406	36	13	17
O	Kaio	199201	200607	85	76	3	728-452	2461	41	2454	35	2441	31	2429	27	0	0
O	Chiyotaikai	199507	200607	63	50	4	555-343	2419	54	2418	49	2413	44	2402	38	0	1
O	Dejima	199609	200607	56	53	4	456-343	2394	66	2373	66	2341	67	2312	71	0	0
O	Miyabiyama	199811	200607	45	43	2	369-282	2392	67	2360	69	2326	74	2298	76	43	32
O	Kotooshu	200405	200607	14	12	0	135-74	2382	70	2332	78	2245	120	2216	130	81	31

The ozeki table is much bigger already with five current plus two former ozeki. Two ozeki are new compared to one year ago with Kotooshu and Hakuho getting promoted meanwhile. Kotooshu had a very promising ozeki run, but injuries prevented him from showing strong performances at the new rank. Right now he would be one of the worst ozeki since 1960, and he isn't exactly showing immediate improvement with his current rating being 75 below his peak. His career still is quite short with just 12 makuuchi basho, so he still can improve his values.The other new ozeki is Hakuho of course, and he didn't stop with the promotion as he is 27-3 since. His peak rating is right now naturally, and only a few ozeki had a better one; Kitanofuji, Mienoumi, Wajima and Wakamisugi in the last ozeki basho and Asahifuji after the three basho 1989.01 until 1989.05 without getting promoted (he got promoted to yokozuna later but never reached this peak again). Even the best ozeki rating of Takanohana II was lower despite winning two zensho in the end as always a bad basho would drag his rating down during his ozeki time. In comparison with the yokozuna since 1960 Hakuho still would be in the bottom quarter right now but as he is quite young I expect him to improve his peak rating in the future.Out of the three current longtime ozeki only Tochiazuma could improve his peak rating with a yusho in the Hatsu basho and elevated his career level to be about equal with Kaio now which is among the best ozeki all-time, but nowhere near yokozuna level. Kaio and Chiyotaikai couldn't even improve the R10 as both slipped over the last year to be currently about 150 (Kaio) and 100 below the peak.No improvement for former ozeki Dejima as well and nothing to expect in the future as his rating is sliding slide from 2204 to 2127 as he enjoys regularly being at the fringes of the joi-jin. It is totally another story for Miyabiyama who surprised many observers with a breakout year as the former ozeki actually improved his peak rating by 43 and also the R10 by 32. Just a year ago he was the worst ozeki since 1960 along with Masuiyama, now he is at least leaving Masuiyama alone in that spot. He still would be quite at the bottom of the ozeki ranking at his age we'll have to see if he can continue to improve.

Sekiwake 1960-2006

Rank	Shikona	First	Last	Basho	Makuuchi	Kyujo	W-L-D	Peak	Rank	Rating 3	Rank	Rating 6	Rank	Rating 10	Rank	Δ P	Δ R10
S	Kotomitsuki	199911	200607	39	34	2	342-227	2365	75	2353	72	2331	72	2311	72	0	11
S	Wakanosato	199711	200607	49	44	4	416-294	2354	81	2346	74	2333	71	2316	70	0	5
S	Tosanoumi	199411	200607	70	65	1	535-510	2317	100	2308	92	2291	90	2276	85	0	0
S	Kyokutenho	199603	200607	62	47	0	451-475	2278	123	2270	120	2254	116	2228	118	0	15
S	Kotonowaka	199007	200511	89	86	4	630-668	2268	136	2246	140	2227	137	2211	133	0	0
S	Takanowaka	199905	200607	42	33	2	304-297	2262	139	2237	146	2209	148	2169	164	0	1
S	Hokutoriki	200201	200607	28	26	0	203-215	2257	144	2196	186	2162	199	2132	214	0	5
S	Tochinonada	199611	200607	58	54	1	421-422	2241	159	2238	144	2233	132	2222	124	0	0
S	Tamanoshima	199909	200607	41	34	0	309-301	2235	169	2233	148	2222	140	2196	142	0	33
S	Tamakasuga	199503	200607	68	55	1	487-502	2225	175	2197	179	2182	169	2164	169	0	0

There isn't much to tell about the bunch of active career sekiwake as most are on the decline. Noone was able to improve his peak rating. Only Tamanoshima came quite close to his peak, improving his R3, R6 and R10 considerably; still on a rather low level for a sekiwake. Others still showing some sign of improvement or Kyokutenho and current sekiwake Kotomitsuki.

Komusubi 1960-2006

Rank	Shikona	First	Last	Basho	Makuuchi	Kyujo	W-L-D	Peak	Rank	Rating 3	Rank	Rating 6	Rank	Rating 10	Rank	Δ P	Δ R10
K	Iwakiyama	200203	200607	27	23	0	206-189	2246	154	2226	155	2205	152	2165	166	0	26
K	Takamisakari	200001	200607	34	29	2	253-242	2241	160	2207	171	2176	183	2148	198	0	9
K	Roho	200401	200607	16	12	0	132-102	2221	181	2208	168	2163	198	2147	200	24	56
K	Kakizoe	200303	200607	21	18	0	160-155	2214	192	2200	178	2168	190	2152	186	0	3
K	Kisenosato	200405	200607	14	11	0	110-98	2205	206	2126	275	2068	353	2055	364	207	103
K	Toki	199707	200603	51	37	1	352-400	2204	208	2183	206	2161	200	2143	204	0	0
K	Asasekiryu	200207	200607	25	21	0	187-164	2198	214	2165	226	2139	232	2100	259	60	39
K	Ama	200403	200607	15	11	0	119-105	2196	217	2120	288	2062	369	2048	381	123	62
K	Shimotori	200105	200607	31	20	1	230-225	2186	228	2153	239	2127	239	2087	295	0	0
K	Chiyotenzan	199701	200511	48	22	1	326-387	2174	242	2151	241	2118	257	2069	327	0	0
K	Kyokushuzan	199503	200607	68	60	0	466-548	2166	253	2165	228	2160	205	2151	190	0	4
K	Futeno	200305	200607	20	14	0	148-144	2166	254	2147	243	2099	293	2062	345	0	46
K	Tochinohana	199901	200607	38	19	1	272-280	2166	255	2097	334	2062	370	2029	430	0	24
K	Kaiho	199705	200607	54	42	2	382-419	2148	274	2134	263	2112	273	2089	293	0	0

We had some komusubi debuts over the last year with Roho, Kisenosato, Asasekiryu and Ama all increasing their peak rating in that period as well. Kisenosato and Ama had the biggest increase as they profited from increasing makuuchi experience and are getting better at a young age. Asasekiryu also had a nice improvement going with the debut, he also still is young getting 25 in 12 days. We will see if the can continue to improve over the next year.None of the older komusubi could improve the peak rating while some padded the R10 with more basho experience. The six komusubi from Shimotori down are in the lower half of the komusubi since 1960, and I see only Futeno as having potential for improvement.

Maegashira 1960-2006

Rank	Shikona	First	Last	Basho	Makuuchi	Kyujo	W-L-D	Peak	Rank	Rating 3	Rank	Rating 6	Rank	Rating 10	Rank	Δ P	Δ R10
M1	Kokkai	200305	200607	20	16	0	160-137	2241	161	2223	157	2184	168	2159	175	0	39
M4	Baruto	200509	200607	4	2	1	47-13	2222	180	2107	311	2092	310	2086	297	New	New
M1	Kotoshogiku	200407	200607	13	9	0	107-88	2167	251	2114	301	2071	350	2061	350	99	37
M3	Tokitsuumi	199705	200607	54	43	2	381-411	2146	282	2142	253	2122	246	2101	256	0	1
M1	Aminishiki	200001	200607	40	35	0	285-304	2145	283	2131	268	2119	255	2098	264	26	21
M1	Tokitenku	200403	200607	15	11	0	118-105	2133	305	2108	310	2062	368	2049	379	95	60
M2	Hakurozan	200409	200607	12	7	0	92-88	2097	372	2048	435	1999	488	1987	493	136	50
M1	Tochisakae	199803	200607	37	24	3	267-256	2097	374	2081	368	2067	356	2045	388	0	1
M3	Takekaze	200209	200607	22	16	2	169-148	2087	394	2084	361	2072	346	2067	333	0	6
M1	Buyuzan	200005	200607	34	25	0	239-260	2080	410	2073	381	2060	374	2036	414	0	5
M5	Toyozakura	199809	200607	30	12	0	210-233	2069	431	2031	467	1998	489	1964	528	0	18
M6	Jumonji	199801	200607	41	31	1	293-318	2067	433	2062	403	2048	409	2034	420	0	11
M6	Kasugao	200207	200607	25	14	0	181-185	2066	437	2046	436	2012	473	1990	488	0	16
M8	Toyonoshima	200405	200607	14	10	0	114-96	2062	443	2044	440	2027	448	2023	445	75	41
M4	Otsukasa	199601	200607	59	25	0	420-465	2047	470	2032	462	2017	466	2002	477	0	0
M10	Ushiomaru	200201	200607	23	7	3	172-163	2038	481	2017	482	1986	502	1961	536	0	-11
M11	Homasho	200601	200607	4	2	0	37-23	2033	488	1986	521	1980	513	1978	505	New	New
M5	Kasuganishiki	199907	200607	31	12	1	215-231	2016	506	2000	503	1985	504	1959	538	0	13
M8	Takanotsuru	200103	200603	29	5	1	200-229	2014	507	1982	527	1952	554	1928	584	0	-6
M8	Tamarikido	199909	200607	27	11	4	187-186	2009	512	2004	499	1986	501	1969	519	0	-5
M11	Yotsukasa	199805	200511	41	8	0	275-324	2003	523	1984	522	1957	544	1932	573	0	0
M9	Yoshikaze	200507	200607	7	4	0	56-48	2000	529	1971	541	1967	534	1965	524	127	92
M9	Kitazakura	199807	200607	43	9	1	314-330	1997	532	1982	528	1960	541	1941	559	0	15
M10	Harunoyama	200011	200607	32	3	2	225-246	1981	546	1954	562	1925	595	1901	614	0	-6
M11	Wakatoba	200105	200607	28	7	0	191-228	1976	553	1963	552	1939	569	1910	603	0	9
M12	Shunketsu	200101	200607	19	5	1	133-150	1972	558	1941	577	1917	605	1910	605	13	1
M9	Tamaasuka	200411	200601	8	3	0	46-49	1972	559	1922	606	1907	613	1903	611	0	1
M13	Katayama	200407	200607	13	5	0	94-101	1960	574	1938	582	1927	592	1924	588	0	15
M9	Kobo	199901	200607	43	4	0	298-333	1948	589	1941	579	1923	600	1907	606	0	1
M16	Daimanazuru	200401	200607	12	1	0	81-98	1888	656	1857	670	1833	685	1828	689	71	34

This is the last table, and I am listing all maegashira this time (as opposed to only those with ten makuuchi basho one year ago). Kokkai still is leading all maegashira since 1960 and he still couldn't get to the komusubi rank during the last year. This could change now as he posted a 10-5 from M5 this basho; ironically his current rating still is about 50 below his peak. Comfortably in second place already with only two makuuchi basho is Baruto, and I don't think he'll finish his career as a maegashira... Kotoshogiku in the next place was severely out of his depth in the joi-jin but I could still fathom him reaching sanyaku in the future.Standing out in last place is Daimanazuru with a peak of 60 lower than that of the next-to-last, Kobo. Still, he isn't the worst all time as Oko (1868), Makimoto (1851) and Azumanishiki (1846) had even lower peaks as moto-maegashira. All of them were in makuuchi only for one basho naturally.

[Asojima 2,870]

Thanks much for the masterwork update.

(Sign of approval) (Sign of approval) (Sign of approval)

[Fay 1,677]

Thank's very much for this great analysis. Well I fear I have to go back to the old forum post and spend an evening with it ;-) to understand it all, but it's worth it.

[Zuikakuyama 1]

A very big thanks for your analysis!!! (Applauding...)

A question.

The data you have here seems to be different than those on Here

Are they different?

[Doitsuyama 1,173]

The data you have here seems to be different than those on Here

Are they different?

The two sets are quite similar. The difference results from two changes I did for the historical ratings. The most important and most obvious change is that this set adds all makuuchi bouts from 1927 to 1988 in addition to the Heisei ratings which leads to calculated ratings where the Heisei ratings had to assume entry ratings for all rikishi in 1989.01. The second and more subtle change is a more refined method to determine entry ratings (= ratings for rikishi entering the rating system for the first time). The Heisei ratings simply have a fixed entry rating which definitely wasn't possible for the historical ratings due to greatly varying size of makuuchi, and especially the Shunjuen incident, so entry ratings are now based on banzuke rank (as a rank number, counting from 1 upwards; eg J14w now would be rank #70).

[Guest Quentooshu]

The data you have here seems to be different than those on Here

Are they different?

The two sets are quite similar. The difference results from two changes I did for the historical ratings. The most important and most obvious change is that this set adds all makuuchi bouts from 1927 to 1988 in addition to the Heisei ratings which leads to calculated ratings where the Heisei ratings had to assume entry ratings for all rikishi in 1989.01. The second and more subtle change is a more refined method to determine entry ratings (= ratings for rikishi entering the rating system for the first time). The Heisei ratings simply have a fixed entry rating which definitely wasn't possible for the historical ratings due to greatly varying size of makuuchi, and especially the Shunjuen incident, so entry ratings are now based on banzuke rank (as a rank number, counting from 1 upwards; eg J14w now would be rank #70).

Incredible work!

What database do you have that contains all of the results and is it available?

Also, does your rating calulator change the ratings of each rikishi after each bout or after each basho?

[Ryukaze 0]

Im a little dissapointed in you bro, seems to me like you plug in this formula and use it to make your sumo game picks eh? (or anyone who uses any type of formula etc.) As they should come from you, your own head and knowledge of watching individual bouts etc. not with the help of statistics, I dont agree.

Edited July 30, 2006 by Ryukaze

[Fay 1,677]

Im a little dissapointed in you bro, seems to me like you plug in this formula and use it to make your sumo game picks eh? (or anyone who uses any type of formula etc.) As they should come from you, your own head and knowledge of watching individual bouts etc. not with the help of statistics, I dont agree.

Hey bro, this is really one of the biggest bollocks I've read in the last time .... (I am not worthy...)

[Zeokage 0]

I agree, the analysis gives a scientific analysis of past results. All that can do for the future is tell you how they're currently doing. For all we know, Futeno could go 13-2 next basho which would be completely against the science. You can predict a trend using science, but it will not be 100% accurate. If it was, there wouldn't be betting.

[sumofan 6]

I agree, the analysis gives a scientific analysis of past results. All that can do for the future is tell you how they're currently doing. For all we know, Futeno could go 13-2 next basho which would be completely against the science. You can predict a trend using science, but it will not be 100% accurate. If it was, there wouldn't be betting.

Indeed, but this is exactly what bookmakers do.

They use statistical analysis to determine the odds, and then use those to calculate the odds they are going to give the gamblers.

Strange things, miracles and disasters happen during the games, but on average, the games turn out as predicted.

That is why bookmakers and casinos always win in the long run.

And I do NOT agree with Ryukaze.

You can make sumo predictions any way you want. flipping coins, using your gut feeling, or statistical analysis.

Why would you not use the knowledge that you have?

That's called improvisation.

If I want to solve a difficult sudoku or logic puzzle, I 'll write an application to do so. Just because I can.

Instead, I am grateful that doytsuma shares this knowledge with us, instead of keeping everything to himself, which he could do as well.

[_the_mind_ 0]

I agree, the analysis gives a scientific analysis of past results. All that can do for the future is tell you how they're currently doing. For all we know, Futeno could go 13-2 next basho which would be completely against the science. You can predict a trend using science, but it will not be 100% accurate. If it was, there wouldn't be betting.

i sincerely doubt Doitsuyama-san went through all the trouble of making this in an attempt to predict the future. it is a ranking of where each rikishi stands in relation to all previous rikishi(and current). it is in no way trying to predict the future. it is a comparison ranking of the past. .....i could be wrong, as i did not recreate it so i dont know its true intention, but that is how it is presented. thats is why he called it "historical strength analysis" not "here is my predictions for the future"

Edited July 30, 2006 by _the_mind_

[Coo-cook 0]

I agree, the analysis gives a scientific analysis of past results. All that can do for the future is tell you how they're currently doing. For all we know, Futeno could go 13-2 next basho which would be completely against the science. You can predict a trend using science, but it will not be 100% accurate. If it was, there wouldn't be betting.

i sincerely doubt Doitsuyama-san went through all the trouble of making this in an attempt to predict the future. it is a ranking of where each rikishi stands in relation to all previous rikishi(and current). it is in no way trying to predict the future. it is a comparison ranking of the past. .....i could be wrong, as i did not recreate it so i dont know its true intention, but that is how it is presented. thats is why he called it "historical strength analysis" not "here is my predictions for the future"

Who knows, we might have a Sumo- Nostradamus among us. (I am not worthy...)

Edited July 31, 2006 by Coo-cook

[Ryukaze 0]

Notice the key words, "It seems to me....." just my thoughts and opinions, (which is what this board is here for right?) I may very well be right with regards to "Doitsuyama-sama's" uses of the information he has (seemingly) painstakingly gone through the trouble to gather, and I may not be. Either way it's just that "my opinion" so "RELAXXXXXXXXX" he's a big boy so he can defend himself if he feels the need to, yet Im sure by now he knows not to take anything I say too personally, he he he

(Incidentaly regarding sumo games however I do indeed still think that the highest degree of skill is shown when someone can predict the outcome of a match using their own personal knowledge having watched bouts etc. as opposed to using any kind of statistics, win % etc to pre-determine the likelyhood of a bouts outcome.)

Edited July 31, 2006 by Ryukaze

[Naganoyama 5,841]

I think it shows a considerable amount of skill to carry out analysis like this also.

Jealous Lollipop?

(Chucking salt...)

[_the_mind_ 0]

(Incidentaly regarding sumo games however I do indeed still think that the highest degree of skill is shown when someone can predict the outcome of a match using their own personal knowledge having watched bouts etc. as opposed to using any kind of statistics, win % etc to pre-determine the likelyhood of a bouts outcome.)

so if Asa were to fight takamisakari you wouldnt already know the outcome? based on the fact you personally know Asa usually only loses one or two a whole basho and sakari is usually scraping just to get 8. and the fact you know the rank difference and that Asa owns everyone except hakuho. you dont think that is using your personal knowledge of the statistics of sumo? it is, it just isnt written on paper. you are using statistics when you try to pick a winner or you would just use a dart board to pick the winner. you can act like you arent using statistics, but you are. its just in your head, the knowledge you have of previous fights, insead of a list.

[sumofan 6]

so if Asa were to fight takamisakari you wouldnt already know the outcome?

Well, if that would ever happen (suppose all sanyaku rikishi go kyujo), Taka might actually amaze you.

He is pretty strong, and most of all, he is unpredictable.

He manages to lose matches that he should win, based on ranking.

He also manages to win against a strong Tochiazuma, who is someone who could win against pretty much all rikishi when he is not injured.

So with Takamiskari, you never know until the fat lady sings.

[Kintamayama 44,349]

Doitsuyama is on holiday at the moment, so he can't answer y'all dudes, bro. (LOL). We have an ongoing rivalry in all sumo games (till Fujisan came along and showed us how it's done..) I always make fun of him that his picks are analytical as opposed to my gut feelings. To each his own, I don't see no problems wif dat..

Edited July 31, 2006 by Kintamayama

[aderechelsea 124]

ok .. since Doitsuyama is not around it is the best opportunity to state that this topic is completely insanish (sic).

i cannot follow it even for a bit.

i guess that's why i never do good in the games when results are "normal". Do you get the same Kintamayama?

The bad thing is that the results in Sumo tend to be very "normal" so that's why Doitsu is so much better than me in the whole gaming world.

I wonder if Fujisan is using his tables or if he is going by gut feeling too, and if he is WHY THE HELL DOES HIS GUTS PREDICT THE FUTURE SO MUCH BETTER THAN MINE ???

Edited July 31, 2006 by aderechelsea

[Jakusotsu 5,806]

Psst, a secret: Fujisan doesn't rely on his own guts - he uses rabbit entrails instead. (Chucking salt...)

[_the_mind_ 0]

ok .. since Doitsuyama is not around it is the best opportunity to state that this topic is completely insanish (sic).

i cannot follow it even for a bit.

i guess that's why i never do good in the games when results are "normal". Do you get the same Kintamayama?

The bad thing is that the results in Sumo tend to be very "normal" so that's why Doitsu is so much better than me in the whole gaming world.

I wonder if Fujisan is using his tables or if he is going by gut feeling too, and if he is WHY THE HELL DOES HIS GUTS PREDICT THE FUTURE SO MUCH BETTER THAN MINE ???

perhaps it might be less confusing if you check out the link in his post to the previous whole ranking. it is pretty well explained. at least i think so.

[aderechelsea 124]

i tried to watch this thread since joining this forum but i guess it is not "my cup of tea" after all. It's probably not difficult but it's not that interesting to me after all ...

anyway .... back to the topic before Doitsuyama returns and finds out i am wrecking his topic .... (Whistling...)

[Kotobaru 0]

Well, back to topic, as a new here it's really amazing to see how many work one :-/ person have been through! My area of knowledge is much related to math and numbers, so they tell me a story. I think this analysis giving the best possibility to compare rikishi of present time to those who are out of dohyo. I hope Doitsu will be back soon to answer my question. I'm really wondering where you got the idea to use the Elo kind of ranking system? Is there any argumentation why it's better than some other, or is it just so easy algoritm to use? Can you give some light ... ;-)

Thanks.

[mokele 3]

These numerical ratings are very similar to those in the chess world, and your work is similar in some ways to that of Jeff Sonas:

http://db.chessmetrics.com/

In any case it's a great idea. I've been a sumo fan since around 1992, diligently watching every basho as long as I could access it. I've noticed that the Kyokai doesn't have a strict or well-defined manner in which the ranking of a maegashira rikishi changes from 1 basho to the next. The only thing that a sumo fan seems to know is that, among maegashira, a rikishi's ranking rises (if there's room for him to rise of course) if he scores kachi-koshi and falls with a make-koshi result, and better scores result in a greater rise in the ranking whereas a very bad result results in a large drop in the ranking. Sekiwake and komusubi will invariably drop out of sanyaku if they win less than 7 bouts in a basho, and will at least keep their ranking if they score kachi-koshi.

1 thing that a numerical rating for rikishi could do is to allow a fairly reliable prediction of how a rikishi will do in a given basho based on his rating and the ratings of his opponents. Of course intangible factors are involved too such as injuries and other health issues. Also, psychological factors seem to play a role, as the plethora of "elevator rikishi" demonstrates. Many of the maegashira rikishi have good results when ranked near the bottom of Makunouchi but have disastrous results when ranked near the top. I suspect that this phenomena is due not just to the strength of the rikishi they face but their mental attitude as well. It's a complicated issue.

Anyway, great research and study! It is appreciated.

[Zeokage 0]

This analysis is brought together using formulae derived from the Elo system, which was originally used for chess...

[Doitsuyama 1,173]

Im a little dissapointed in you bro, seems to me like you plug in this formula and use it to make your sumo game picks eh? (or anyone who uses any type of formula etc.) As they should come from you, your own head and knowledge of watching individual bouts etc. not with the help of statistics, I dont agree.

I certainly use my statistics as a help for making game picks, not solely by any means, and not even exclusively as anyone is invited to do the same (the Heisei ratings are freely available after all).

But even if I would use this stuff solely and exclusively - I don't care if you agree or not. I still feel it's a valid approach.