Tuesday, 20 November 2012

JFK Statfest And Comparing Different Ultra Finish Times

9 miles into the JFK 50 course on the AT

Having just got back from the 50th annual JFK50 (see a great history of the race here), it was an honor to be part of the event, especially with course records being destroyed (see results). It got me thinking about the comparability of race times between road and trail races, plus different distances.

On a training run the day before the race it's easy to bump into half the Montrail North American team and a whole bunch of Canadian speedsters.

For interest, here’s the JFK course profile which includes around 3,000ft of ascent and a little more descent. It has roads or canal path easy trails for all but 11 miles, of which there’s a really fun 10-mile section on the Appalachian Trail which is quick and generally not technical except for maybe 3-4 miles of rockiness. I may write up a report of the race, but there are a few of those around from others who had more interesting days out on the trail. In summary, I loved the AT and jogged through that then sped up on the canal path before getting slight stomach problems (rare for me) and couldn't get motoring so just kept up the best pace I could and mainly ran solo, finishing 4th in 5:50.

Predicting finish times from one ultra to another

One stat I’d heard at the Miwok 100k in California is that if you double your time from that race you get your Western States 100 mile finish time. Although broadly true there’s a lot of variation, plus that assumes having a good day at both races. It’s glaringly obvious that there’s a large degree of individual impact on how well, say, a marathon time translates to a flat ultra and even more so to a mountainous one. But JFK brought some of the fastest runners ever to a US trail 50-miler – Trent Briney ran a 2:12 marathon and has been the US alternate for the Olympic marathon (he got 2nd and broke the CR); Max King runs a 2:14 and took almost 6 mins off the record; Emily Harrison runs a 2:32 and smashed the women’s record, but not by as much as 2:42 marathoner Ellie Greenwood; plus there were a lot more guys with 2:2x marathons, not least David Riddle who came 3rd and held the record from last year.

Ultras like Comrades, UROC 100k, JFK 50 and American River 50 have plenty of fast road sections (or trails that are almost as quick) so they favor a fast marathoner who also trains for ultras. But there’s a lot more that comes into it too (as anyone reading this will certainly know). A quick comparison of some of the fastest ultra courses for people who’ve run them at the front shows road speed is fairly important, but more so the shorter the race. I’m biasing this towards courses I’ve personally done so I can legitimately compare them.

Comparison of PRs* for selected runners over multiple fast ultra courses:

Marathon (26.2)
JFK (50.2)
American River (50)
Comrades Down Run   (55.5)
UROC (60+** 2012 course)
100k Roads (62.2)
Rocky Raccoon (100)
Western States (100.2***)
Eric Clifton
6:23 (Age 40)
Not raced
Max King
6:01 (Bad day)
Not raced
No 100k
No 100s
DNF (injured)
Trent Briney
2:12 (2004, but ran a 2:19 in 2011)
Not raced
Not raced
Not raced
Not raced
Not raced
Not raced
David Riddle
Not raced
Not raced
DNF (Bad day)
Not raced
Ian Sharman
6:00 (Bad day)
Not raced
Ann Trason
Not raced
Not raced
Not raced
Ellie Greenwood
Not raced
Emily Harrison
Not raced
Not raced
Not raced
Not raced
Not raced
Not raced
Lizzy Hawker
2:47 (2007)
Not raced
Not raced
6:48 (Up Run - slower course)
Not raced
Not raced
18:32 (Bad day)

*PRs are to the best of my knowledge, plus the help of Google
**UROC 100k 2012 had 10,000ft+ of ascent but it was largely on roads and hugely favored fast marathoners who could also run trails
***Western States 100 isn’t nearly as flat and fast as the other races but it’s the only 100 miler for some of the athletes to date

Before I get comments saying this isn’t a large enough sample or it’s not scientific enough, I’ll mention that there aren’t enough runners who’ve run around course records at several of these races to make direct comparisons plus weather conditions play a huge part, especially on trails. More runners could have been included but I’m doing this with a JFK 50 2012 slant and am trying to show the really pointy end of the field, (plus myself for comparison because it’s my blog and it gives me context).

Along the C&O Canal at JFK. Photo courtesy Ray Jackson Jr.

Also, I'll point out that a great site for comparing race finish time for the entire field from race to race is RealEndurance.com. You can get an idea of your potential finish time in a race based on what other people managed at a given event who run at your speed in races you've already done. 

What does that imply for 100 milers?

The other reason for looking at these stats is that I’m sure a lot of people are wondering what Max, Trent or Emily could run in a 100 miler. From experience the correlation between fast shorter distance times and 100s is a lot looser than between a marathon time and a race like JFK. Plus there’s a clear trade-off between how much time a runner can spend on road speed and on mountain endurance, although most would agree that these do complement to some degree. Kilian isn’t going to run a 2:10 marathon off pure mountain running and the Kenyans won’t run a 20 hour Hardrock 100 when their longest run is 2 hours and at a much higher intensity. Of that I’m certain, although it’ll never be tested (at the least they’d switch their training significantly if they went for the races at the opposite end of the spectrum to their usual).

To run an average pace of 9 min/miles at Western States means a lot more training at a slower speed than a marathoner would do. This is due to the law of specificity, meaning that your body adapts to the training stresses that are placed on it. If you want to run a fast marathon, a lot of marathon-pace or faster running is required. If you want to sail up and down mountains all day long, a large portion of training needs to simulate that.

When was the last time one of the really big, competitive 100 milers was won by a sub 2:30 marathoner? Doesn’t tend to happen at Western States (please comment if you know who the last person was to manage this was as it’s not any of the recent winners). UTMB is for pure mountain guys and many of the top runners haven’t even run a road marathon. Never mind Hardrock – a recent fast marathon time is almost (I’m exaggerating) a predictor of a bad run as it implies too much time spent on the roads and not enough in the mountains at altitude. It’s much more important to do a lot of vertical in training than to be able to run the flatter sections at a 5 min/mile.

In summary, I don’t think there’s a very strong relationship between mountain ultra success and a top end marathon time. I’m not going to dust off the old economist’s tools (my previous life) and search for a huge pile of data to find out which variables correlate to mountain ultra success. It’d cost $20k to get a bunch of economists to do that analysis so I’m going to go out on a limb and give you my non-scientific predictions of the key variables for a fast elite time (relative to a world class runner for that style of race) at a given mountain 100-miler:
  1. Results at really similar 100 mile races or with similar aspects to the race in question
  2. Turning up completely uninjured with a long injury-free period pre-race for consistency in training
  3. Location – living close to terrain that’s similar to the race for training
  4. Frequency of DNFs – the fewer, the better
  5. Motivations – this’d be a hard one to model and would need truly honest answers to a questionnaire but someone who turns up with the aim of enjoying things first and competing second rather than caring more about records and winning with second place being deemed a ‘failure’

And factors that I think are somewhat correlated:
  1. Past success at the race in question in the recent past, but this could also build the pressure too much to cause bad pacing
  2. 100k or 50 mile mountain results
  3. Marathon time
  4. Age – Marco Olmo is probably the last guy around retirement to win a really major race but the peak age range is fairly wide
  5. Rippling six-pack/big guns – shows the runner does more training than just running all day, which is more important in an ultra than in shorter races. You wouldn’t pick a fight with Kami Semick and she’ll probably beat you in the race too…

100 mile records

So what about a flat 100-miler on a hard surface? There really aren’t many that fit that description except on a track. Fast trail 100s like Rocky Raccoon 100 (5,400ft of ascent) or Umstead 100 (8,000ft of ascent) are still significantly slower than a flat road race of that distance. What do I think the runners listed above could run on a flat road/track 100? Well, the 100k on roads gives an indicator but only covers the ‘easy’ bit of the race.

The male World Record for 100 miles is 11:28 by Oleg Kharitanov, pretty much 3h marathon pace x4! I think Max and Trent have the pace to do that if they altered their training to include a lot more miles at a pace they’d think of as slow.

Ann Trason holds the female World Record of 13:47 which I think Ellie could run. Lizzie Hawker too.

But who’d want to run around a track all day? Well, I’ll answer that in a month after I give it a go at the Desert Solstice 24h race in Arizona.


  1. I agree that there is most likely little correlation between marathon time and mountain 100 mile times. But there does appear to be fairly strong correlation between times in mountain ultras of 50 miles or less and marathon time.

  2. Ian,
    I always agree 100% with your analyses. Great job making people think!
    Wanted to get your feedback on this "rule of thumb": In this HADD method to distance running the optimal runner is said to have 15 second intervals between their 5k, 10k, half and marathon times respectively...so your 5k time would be 1:00 less than your marathon per mile pace. http://www.angio.net/personal/run/hadd.pdf

    Do you think you could come up with a corresponding "clock" for 50k, 50 mile, 100k and 100 mile on runnable singletrack? I think a per mile pace table of the above runners might highlight correlation more than finish time. Also add in Wardian and Mackey.

    Great analysis.

  3. James - I think the marathon/ultra calculator's give a good idea of the average transferability of, say, a 5k race time to a marathon time, but there's a huge amount of variation. For example, on McMillan's Running Calculator my marathon time gives me a 5k time that's over a minute faster than I can actually do. The correlation with long stuff (beyond a 50k, which is basically a long marathon) is even weaker since a lot more comes into play than just cardio fitness. McMillan suggests my 100 mile time is equivalent to a marathon WR!

    So the way I think about it is that everybody has an individual curve of the percentage they slow down per mile as their race distances increase on flat/easy terrain. But the angle of that curve isn't nearly set in stone for all humans, which is what the conversion calculators assume, using an average. But all other things being equal, if your time is improved at one distance it should improve at other distances...that assumes the training involved for that faster time didn't involve a trade-off in specific training for the other distances, which is not always possible.

  4. Ian,

    One guy with a pretty decent time in a marathon (2:38) and 7-time WS winner is Scott Jurek. Now I'm not sure which came first, and he is definitely not a sub-2:30 guy, but it shows that it can be done.


  5. John,

    2:38 isn't a fast marathoner for men (Ann Trason ran a 2:39), being a min/mile slower than the likes of Max/Trent. Most of the top ultra guys can run a 2:3x and some a 2:2x, but I can't think of any 2:1x guys recently who've run major 100 milers and done so at the very top level (I can't even think of any who've been mid-pack, but that's partly due to there not being many guys that fast and that they don't tend to run 100s). But I'd still argue that they don't have nearly the same advantage, if any, as they do in a flatter 50k or 50 miler.

    Also, the 2 speedy marathoners who're tearing up hillier mid-distance ultras, Max and Sage Canaday, have a very important additional skill - they're on the US mountain running team.

  6. This comment has been removed by the author.

  7. Ian,

    I too am wondering all of these questions. As athletes we need challenge and Western States and 100k are on my radar for next year as well as some marathon races. It will be interesting as ultra running's popularity continues to grow how this data will look in 5 years-10 years.

    I think as runners we enjoy the trails and running with other like minded people. I'm hoping to stay healthy this next year and go after some new race distances. I also think its best to keep changing the stressors on the body which helps create adaptation and fitness in the body. It also helps with the boredom of years and years of doing the same thing. Freshen it up! And, of course go run in the hills with great views!

    Really enjoyed reading your blog and think your thoughts are quite accurate. Best of luck in your 24 hour in Arizona!

    Cheers man,


  8. Awesome post. This beats the heck out of the play by play race report.

  9. Ian there is some really good and quite academic research on the correlation of running times vs distance vs VO2max etc in Tim Noakes' 'Lore of Running'. That is mainly around road races, which have the simplification of largely removing the weather and terrain variables, but interesting you have tried to extend it to trail ultras. Good one.