Ultramarathoner PRs: A Look Across the Miles

In my previous life I was an economist, before running took over and I switched to coaching. Rather than the macro-level factors leading to the recent recession (I swear it was nothing to do with me), I dealt with analysis at the micro-level and I have to admit to a lingering love for stats. So from now on, I’ll be writing for iRunFar with my ultrarunning geek hat on. I hope to shed a little light on different aspects of the sport with numbers and analysis.

To start with, the topic of faster road runners coming to ultrarunning has led to more competition and lowered course records. The sample of elite runners included below is small, mainly North American and contemporary with their personal records being more for interest than analysis. Like most runners, this group has focused to a greater level on certain distances than others so some of their PRs are relatively more impressive. Yet I think the numbers alone are fascinating and show the relative strengths of different runners for road or track events. Plus, it gives a rough idea of a range of how much any well-trained (but not necessarily fast) runner’s pace is likely to drop off as the distance increases on the flat, by comparing a runner’s PR at any distance to their marathon time.

These ranges are fairly narrow for shorter distances, but expand more for the ultra distances, partly reflecting that runners can’t be the best at every distance and will be suited to certain distances more than others. If you compare your own PRs in the same way, then you can see how you stack up against the relative strengths of these top-level runners.

A good example is to look at Mike Wardian, a man who has excelled at numerous distances and who races very frequently. As fast as his ultra times are, his shorter-distance speed suggests he could lower his road ultra times because his ratios are relatively high for ultras. That’s something his competitors should worry about!

Note that I selected runners who have at least some road or track background because those types of courses are more comparable than for trail PRs. I left blank PRs that either don’t exist for a runner or that are not at least reasonably reflective of a well-trained effort (by their own admission). The trail PRs reflect choice of race more than ability because a 50 miler like JFK (Max King’s best) is obviously quicker than the hilly Lake Sonoma course (Sage Canaday’s fastest 50). I’ve also included the world-record holders at 50 miles, Bruce Fordyce and Ann Trason–two of the most talented and influential ultrarunners ever.

We’ve embedded the table below. You can also access the data as an image or Excel file.

  • All data from the athletes themselves except Ann Trason’s data from ultralegends.com and realendurance.com.
  • Numbers highlighted in red have caveats or are adjusted from a close distance (eg. 1,500m extrapolated to 1 mile).
  • Numbers highlighted in purple are estimates for the sake of comparison since Geoff and Rory haven’t run a road marathon.

Call for Comments (from Bryon)

  • What do you glean from the data Ian provides?
  • For these speedsters, where do you see the fastest relative times coming from: genetic predisposition, training focus on a particular distance, relative number of times and times span over which a runner raced a distance, something else?
  • What do your splits across the distances looks like? If you take a look at your relative times, what is the root cause for your fastest relative performances ?
  • While most of the shorter-distance times are likely to have been run on reasonably similar terrain, how do we tease out meaningful trends when an ultrarunner may have only run slow courses or under slow conditions at one of more of these distances?

There are 4 comments

  1. JordanLaF

    Thank you Ian. I geek out on stats like these as well :) I'm newer to Ultra Running so I try to study race times elevation charts and any other piece of data I can get my hands on. I didn't see your name on that list though, would have been interesting…

  2. Emir

    I think it will be hard to compare one race to another with different terrain and elevations. For the most part, road races are pretty similar, but on the trail it is completely different. As more and more of these fast guys get into the races, we will need to compare each race year to year to really see the difference. Great post Ian. Keep it up.

  3. jkallen

    This is pretty cool- thanks, Ian. There's definitely a larger degree of variation at longer races due to bigger variation in terrain and smaller sample- most of us have done enough 5k's to have some degree of confidence, but not so much for 100k or 100m. Plus it's hard to compare anything other than flat races. For example, all my 50 mile and 100k's have been technical races, so my PR's for those distances are splits from my 100 mile PR.
    I'll be the first one to throw up my data: X, 16:01, 33:00, 1:11:37, 2:34:16, X, 3:58, X, 7:03, X, 9:02, X, 15:19, X.

    Bottom line for ultras- you can see general trends of how fast someone is, but can't read too much into the specific data points. The only fair comparison is data from the same race each year, with maybe some comparison year-to-year.

    Thanks for the time spent compiling this, Ian.

  4. AtomLawrence

    It would be neat if there were some way to mathematically treat the effect of elevation gain on speed. Strava's Grade Adjusted Pace is an attempt at this, and is derived from scientific work which I've referenced on the "Speed Comes to Ultrarunning" on track and road ultras here on IRF, but there are obvious problems with this. One is that it doesn't take into account technicality, which is especially slowing on descents. Another is that it deals solely with metabolic output, which (as the scientists noted) is the primary determinant of speed on flats and uphills, but is but one of several constraining factors on downhills (otherwise, everyone would make up the time they lose on the uphills on the descents, and any course with equivalent gain and loss would be effectively "flat" over any meaningful distance). Finally, it seems that how vertical is proportioned throughout a course will have a distinct effect on pace: a summit assault where one climbs 10k ft in the first 25 miles, then descends the same in the second 10k is probably going to be slower than a rolling 50 like Sonoma, with the same vertical distributed much more evenly over the total distance (I could be wrong here, but it seems certain there would be some difference).

  5. bazzzzzzzzzzzzz

    This does throw up one of the interesting aspects of trail running which is how do I judge my own performances in a race? With a road marathon, I can say with some certainty whether a given race was a triumph or a disaster; usually these two extremes are separated by a difference of about three minutes. I'm sure these ridiculously fine balances are one reason why many people find themselves drawn to trail races where the parameters defining success and failure are much more nebulous.

    So when someone asks me after running a trail ultra if I had a good race, I'm not always sure what to say. If I didn't shit my pants or lose any limbs, then the answer is usually yes. If the finish line beer is of sufficient quality then it's an emphatic yes. Sure, for the elites, a good race is defined by winning, and for the novices, it's a question of actually finishing and beating the cut-offs. But for the bulk of us mid-pack hoi polloi, it's not always so clear cut. ultrasignup.com percentages go some way to indicating performance, but only in races where the top dogs show up and are forced to put in a top dog performance. Once you've run a few local races, friendly rivalries start to build up, and these often seem to me to be more of a motivating factor than some vague time goal.

  6. _RobHoughton

    This data is very useful but might I suggest an enhancement to aid comparability and also examine where each athlete is strongest or had their best race. If you collect the fastest known times for each distance and then divide the fastest known time by the individual runners time then you have a percentage relative performance. This lets you see the distance where that runner has had their best performance and how they compare against others. This also allows comparisons of men versus women and is something mere mortals like me can also do to see what percentage or our hero's we are currently performing at. I will start you off by giving the fastest times for mile, 10k (not 10,000m), half marathon and marathon which are 3:43, 26:44, 58:23 and 2:03:23 for the men and 4:13, 30:21, 1:05:27 and 2:15:25 for women._Enjoy and what do you use to compare your performances?

Post Your Thoughts