Athlete Country 5000m 10,000m Half Marathon Marathon
Haile Gebrselassie Ethiopia 12:39.36 26:22.75 58:55 2:03:59
Paul Tergat Kenya 12:49.87 26:27.85 59:17 2:04:55
Zersenay Tadese Eritrea 12:59.27 26:37.25 58:23 --
Moses Mosop Kenya 12:54.46 26:49.55 59:20 --
Ritz will be the next one added to this list. Mark it!
Ritz still has to run 2/3 on the list. I think guys like Sihine or Gebremariam are closer. Dinkesa will be the next though. He runs a good bit on the roads.
Since I'm not doing anything right now, I just took those times and distances and plotted them for Geb and Tergat, imposed a linear trendline to get slopes. Oddly they line up very well on a linear fit. Of course, y intercepts are negative using this model, so it's not perfect. Slope for Geb 3.0036, for Tergat 3.026 (units would be min/km). Y intercepts -3.3117 for Geb, -3.3596 for Tergat.
I took the times for Tadesse and plotted those, too, with a slope of 2.8266 min/km, and y-intercept of -1.3494. Whoa, much flatter slope across the distances. Using this VERY inaccurate method to predict his marathon time gives a 1:57:56! HOLY SHIT!
Let me fit these with some kind of polynomial that has a 0 y-intercept and I'll get back with maybe some more realistic predictions.
Until ventolin weighs in on this, your figgers are on a slippery slope, I'd say.
OK, using a power based regression, which fits even better, and of course has 0 Y-intercept, I come up with fit equations for Geb of y=2.2524x^1.0706, for Tergat of y=2.283x^1.0684, and for Tadesse of y=2.4151x^1.044. This gives a predicted marathon time for Tadesse of 2:00:09! HOLY SHIT.
All this is of course, based on the idea that Ventolin type predictors work, where an athlete's predisposition toward longer events causes a flatter slope across performances and will predict faster performances out on the long end.
His 5000m is comparatively slow. Don't think BK (not KB!)
was ever in sub-27 shape...
Yeah, we have to assume that his 12:59 5000m is the best he could do. Running a 26:37 would suggest that he could run much faster. But then.....maybe not that much faster. Maybe his slope is that flat. After all, look at Gebs faster 5000 and 10000m times and comparatively slower HM time.
This is all fun n games. don't take this seriously
Logically Bekele would be a slam dunk if he'd ever try one. Apparently Sihine has done at least 1 half and went 61:14. I'm sure there are several sub 13/27 Kenyans already too.
i'm never too propulsive in advocating calcs for 1/2M & M based on track times as former doesn't have benefit of relative windless conditions ( in modern enclosed stadia ) & wabbits/elites who can gauge/adjust adjust their pace every 400m by looking at stadia clock ( it can even give them projections ! )
trying
all we can guess is that tadesse is likely in something like
12'40 - 12'50 / 26'20 - 26'30
current shape ( watch his races over 10k especially osaka when he almost squeezed the life out of kennster - tadesse is definitly close to 10k wr ability & if you can do that, you need to have ``12'50 speed )
trying the variations :
12'40 / 26'30 ->50.6 , 1'46.8 , 3'29.8 , 7'20.6 , 58'34 , 2"02'03
( that line is similar to komen - for track )
12'45 / 26'30 ->52.5 , 1'49.8 , 3'33.4 , 7'25.8 , 58'10 , 2"00'35
12'50 / 26'30 ->54.4 , 1'52.7 , 3'38.0 , 7'31.0 , 57'49 , 1"59'08
12'40 / 26'20 ->52.1 , 1'49.0 , 3'32.4 , 7'22.8 , 57'50 , 1"59'53
12'45 / 26'20 ->54.0 , 1'51.9 , 3'36.5 , 7'28.0 , 57'28 , 1"58'26
12'50 / 26'30 ->55.8 , 1'54.8 , 3'40.6 , 7'33.2 , 57'04 , 1"56'58
take your pick...
"watch his races over 10k especially osaka when he almost squeezed the life out of kennster"
LOL!!!!!! This is why I enjoy ventolin's posts. That cracked me up.
ventolin^2 wrote:
12'40 / 26'30 ->50.6 , 1'46.8 , 3'29.8 , 7'20.6 , 58'34 , 2"02'03
( that line is similar to komen - for track )
12'45 / 26'30 ->52.5 , 1'49.8 , 3'33.4 , 7'25.8 , 58'10 , 2"00'35
12'50 / 26'30 ->54.4 , 1'52.7 , 3'38.0 , 7'31.0 , 57'49 , 1"59'08
12'40 / 26'20 ->52.1 , 1'49.0 , 3'32.4 , 7'22.8 , 57'50 , 1"59'53
12'45 / 26'20 ->54.0 , 1'51.9 , 3'36.5 , 7'28.0 , 57'28 , 1"58'26
12'50 / 26'30 ->55.8 , 1'54.8 , 3'40.6 , 7'33.2 , 57'04 , 1"56'58
take your pick...
Tadesse ran a workout of 3 x 5 x 500 m in 1:09-1:10 with 2 min recovery and 4 min between sets. That is 55-56 pace per 400m. I think he can run under 52 seconds for 400m and under 1:52 for 800m. No one has the endurance to run only 55.8 and 1:54.8/3:40.6/7:33.2
I think he is capable of under 2:02, but 1:56-1:59 is absurd.
However, I also feel like your formulas don't take into account the additional slowing that occurs in the marathon due to needing to use fat as a fuel source.
cheez-its are nom nom nommm
Obviously we're taking this with a grain of salt, but any prediction which yields an easily sub 2 marathon can probably just be thrown out. I mean, maybe? If EVERYTHING went exactly right, but when has this ever happened?
The challenge here is that it is hard to be fit for multiple distances at once. I think it was Ritz who said last year that to be in peak marathon shape means you have to lose something for the half. Wanjiru, who I agree with the Wejos is probably the best marathon racer right now, usually loses his half-marathons shortly before marathon victories.
My best guess is that Wanjiru could be around 2:03-2:04. At lifetime PRs of 26:40 and 58:33, he's very close to Tadese. I know 10 seconds is a big chunk to tear off, but it's really not very much. It might represent better fitness, and it might just represent a slightly better night's sleep on ZT's part. I realize that Geb's excuses often seem like BS, but there's something to it. I think that it takes an amazing combination of genetics and hard work to get in WR shape, but actually executing on a given day seems fairly unpredictable. There's just so much going on. Also, I doubt that Wanjiru could run 26:40 on the day of his best marathon. Training requires tradeoffs; I discover this as I prepare for the 800 outdoors and even my mile strength is slipping.
Having said that, I was very impressed by Tadese's very close and very competitive performances in the 10k and HM world champs this year. I think he's the real deal. But I'd be absolutely shocked to see him smash the WR. Still, the predictions are fun. It's nice to see y'all use stat methods. I think eventually this kind of analysis will yield relatively accurate predictions.
I mean if a 35 year old man could run 2:03:59 it is a record that has some room to go down. Take a spectacular and "in-their-prime" 10,000/HM talent and have them seriously train for that WR and run at Dubai. I think it goes down. Geb is a great time-trialer. So are Tadesse and Bekele. Wanjiru seems to be more of a racer, but you put him with one of those guys he might win out and run a blazing time.
Poker I will politely disagree about Bekele being a time trial runner. He has run amazing times in settings with rabbits but he has also won some serious hardware from all types of races. Most notably he has his blazing kick from a lap out, but remember in Beijing in the 5000m when he went through his last 2k in damn near 5 flat and the last mile was close to 4. Bekele has talent and fitness that no one else has and can pretty much race anyway he wants so we might always think that he either runs a time trial or blazes his last quarter. But if you look over his results you will see that his last 2k is where he makes people hurt.
Absolutely. KB, ZT and HG all have numerous wins in head-to-head races as well as settings that could be described as time trials. It's hardly worth digging out links to videos, since there are so many.
The Stache wrote:
OK, using a power based regression, which fits even better, and of course has 0 Y-intercept, I come up with fit equations for Geb of y=2.2524x^1.0706, for Tergat of y=2.283x^1.0684, and for Tadesse of y=2.4151x^1.044. This gives a predicted marathon time for Tadesse of 2:00:09! HOLY SHIT.
All this is of course, based on the idea that Ventolin type predictors work, where an athlete's predisposition toward longer events causes a flatter slope across performances and will predict faster performances out on the long end.
Hey Stache...your a big nerd! I totally thought of doing the same thing! But I figured Letsrunners (mean age 17?) would not understand what I was talking about. Do you have a stats background?
the reality of it wrote:Tadesse ran a workout of 3 x 5 x 500 m in 1:09-1:10 with 2 min recovery and 4 min between sets. That is 55-56 pace per 400m. I think he can run under 52 seconds for 400m and under 1:52 for 800m
52.0 / 1'52.0 gets you nowhere,
->3'43.6 , 7'57.3 , 13'53 , 29'25 , 1"05'50 , 2"18'42
No one has the endurance to run only 55.8 and 1:54.8/3:40.6/7:33.2
exactly
it means 12'50 & 26'30 are mismatched times & therefore any guy who can run 26'30 must have current 5k potential far quicker than 12'50
I think he is capable of under 2:02, but 1:56-1:59 is absurd
those long distances are assumed to be run on the track same as the shorter distances
i'd have no hesitation in offering ~ a coupla minutes difference between tunning a M on a road & on a track
However, I also feel like your formulas don't take into account the additional slowing that occurs in the marathon due to needing to use fat as a fuel source.
a 2"15 female has shown that is bogus
Sammy Wanjiru & Bernard Kipyego are the only others to go Sub 27 & Sub 1 Hour. They're both still 23. They probably don't care but it would be nice if they'd take a crack at Sub 13. Sammy has a 13:12 to his credit & Bernard has a 13:09 to his.