Pace charts for TNF 50. And how everyone starts too fast.

The two weeks before a big race can be challenging in the enforced inactivity. Between mid-October and mid-November I ran about 14 hours per week on average, but during my taper for The North Face Endurance Challenge 50 Mile Championship this weekend I’m running less than half that, creating a lot of free time. However, I have found it to be a good opportunity to finalize race weekend details: air and ground travel, lodging, gear selection, aid station or crew support strategy, etc. Last night I worked on pace charts, and the results are interesting enough that I thought I would share them.

A pace chart for the TNF 50 Mile course — and every other mountainous trail ultra, for that matter — is unique to this course. Whereas a pace chart for a flat road marathon at sea level is applicable to hundreds of other marathons, there are too many variables on the course that affect a runner’s pace, notably the cumulative vertical gain/loss and the specific locations of those climbs/descents, trail quality, and altitude. Besides, due to the topography of a mountain course, aid stations and checkpoints are situated where it is most logistically convenient and almost never coincide with mile markers.

Everyone starts too fast, especially the runners who don’t win

In order to develop a pace chart, I needed to find a pace from a past race that I liked. I found seven athletes on Strava who posted on Strava their GPS watch data from their 2013 performance (thanks guys). Since my goal time is 7 hours 20 minutes, I tried to find athletes in this range; I also included some of the top runners out of curiosity. The runners were:

Here were there splits:

Splits of 7 runners from the 2013 TNF 50 Mile race. Data manually retrieved from Strava.

Splits of 7 runners from the 2013 TNF 50 Mile race. Data manually retrieved from Strava.

When I calculated their average pace over the race, it confirmed what is commonly said about this race: the pace starts off too aggressively. For all of the runners, their average pace gradually increased. And the increase was was not equal across the board: Krar’s pace only increased 40 seconds, Dylan’s 60 seconds, and the remaining 5 runners by 85-103 seconds.


A chart is below showing their average pace over the race. My apologies that the Y-axis is in a numerical format that makes hardly any sense (fraction of an hour), but I struggled to create the chart if that series was in a duration format (HH:MM:SS). The point is the same: everyone slows down.


If the course became increasingly harder (i.e. more vertical per mile, poorer trail quality), the significant increase in pace would be understandable. But that’s not the case — while the second half of the course is probably slower, it shouldn’t be 90 seconds slower.


The “pace creep” was more dramatic with the runners who didn’t finish at the top. Consider that the gap between the fastest and slowest pace between the start line and McKennan Gulch (mi 22.7) is about 60 seconds, but it increases to 100 seconds by the end of the race.


Pace charts based on Dylan Bowman

When I began this exercise, I thought that I would probably base my pace charts on a racer who finished around my goal time of 7 hours 20 minutes. But such a pace would have me going out too hard, losing an increasing amount of time per mile to the leaders. I wanted a pace that was more consistent.

Rob Krar ran the most consistent race, but I’m very doubtful that I could match (even proportionally) his legendary surge last year at Mile 40. He was the only runner (at least among those whose data I used) who averaged a faster pace for Miles 40-50 than Miles 30-40. I’m a decent runner, but I’m not in his class.

Of those I analyzed, I like Dylan Bowman’s pace the best. He ran a very consistent race, not starting too fast and then finishing strong. When I proportionally scaled his splits to splits that are more realistic for me, I got these pace charts:


What do you make of my analysis?

Posted in on December 2, 2014


  1. Jamie on December 2, 2014 at 12:00 pm

    What’s the basis for the assumption that positive splits means you started out too fast? It is at least theoretically possible that positive splits can produce the fastest overall time. This would be so if, for example, in order to achieve even or negative splits, you have to start out so slow that you erase any benefit gained by not slowing down later.

    • Andrew Skurka on December 2, 2014 at 12:15 pm

      Fair point and I don’t have the data to counter it (or the know-how even if I did). Indeed, sometimes it makes sense to go out hard in order be in position to finish fast overall. Negative splits usually only happen with heroic efforts or in championship races when everyone goofs around until near the end when the hammer comes down, e.g. this year’s NCAA XC Championships.

      But the spreads between the runners’ early and later paces is pretty dramatic: 90 seconds per mile for 5 of them, 60 seconds for Bowman, and just 40 seconds for Krar. Bowman and Krar appear to have run more within their ability early in the race and therefore were able to close faster; relatively speaking, the other runners got too caught up in the start of the race.

      • GZ on December 2, 2014 at 2:05 pm

        Yes – I have wondered about this same point and if 100 milers are more like the 800 meters – the best times don’t come off a negative split, but that slow down the least.

        I am only a one time 100 guy but folks told me I started to fast. If it was true, the question became, what pace should I start at.

        For example … if you are going to run a 20 hour 100, pretty nobody would suggest you start at 12 min/ mile. So what do you start at assuming a slow down? 11 min/mile? 10 min/mile?

        (not expecting you to know but it is something I have pondered … particularly since it is probably impossible to replicate such a thing in training for a 100).

        FWIW, it gets me wondering if the “rule” that a negative split anywhere is the best approach for PRing at any distance. But that is probably a can of worms.

  2. Wes on December 2, 2014 at 12:21 pm

    Different event, different circumstances, but an example of how it is possible to run for 6+ hours with even splits if you pick the right pace from the start and you are super fit. Max ran a 3:12 50k followed by a 3:15:

  3. Andrew Skurka on December 2, 2014 at 12:21 pm

    I asked Brett Rivers, for whom this course is his backyard, whether the “pace creep” could be explained fully by the course. His response:

    Yes, guys go out way too hard there and A LOT of guys will blow up. The flip side to keep in mind though is that most everyone is naturally going to slow down simply due to the distance and vertical. You could start of super conservative yet still slow down towards the end as the body fatigues. Runners that find the sweet spot will do the best overall.

  4. Katherine on December 2, 2014 at 12:37 pm

    No matter how you do in the race, you should get a special award for the best training analysis!

  5. Speedgoat Karl on December 2, 2014 at 2:24 pm

    Killer read Andrew…and I don’t read many. 🙂 I found it most interesting when you mentioned, “But the spreads between the runners’ early and later paces is pretty dramatic: 90 seconds per mile for 5 of them, 60 seconds for Bowman, and just 40 seconds for Krar. Bowman and Krar appear to have run more within their ability early in the race and therefore were able to close faster; relatively speaking, the other runners got too caught up in the start of the race.”

    I’ve always said we gain more time relative to the field in the latter stages of 100s if we stay well within our comfort zone early. You just confirmed it via math for a 50 miler.

    Next up, I’d love to see the numbers if you could put one of these together from say, WS, where lots of competitive runners role.

    Great stuff, good luck Saturday.

    • Andrew Skurka on December 2, 2014 at 2:38 pm

      Welcome, Karl!

      You have WAY more ultra running experience than I do, so I’m glad to hear you say that you can gain time on the field in the latter stages if you are conservative early. But that takes a lot of discipline — an 8-min pace feels SO slow, and it’s also really hard to watch the lead runners gap you immediately and significantly because they start at 7-minute pace.

      If I’m bored this winter I’ll look into the Western States data. Though after this post I wonder if a bunch of runners will block me from their Strava accounts.

  6. Kircher on December 2, 2014 at 3:46 pm
  7. Dylan Bowman on December 2, 2014 at 7:32 pm

    Good stuff, Andy! I also managed to get off course for ~3mins between Cardiac 2 and Old Inn. Keep your eyes open there. Hope to meet you this weekend!

  8. Daily News, Wed, Dec 3 on December 3, 2014 at 6:53 am

    […] Skurka puts together an awesome analysis of elite Strava data from last year’s TNF.  His findings: Most of the elites go out too fast […]

  9. Steve Collier on December 3, 2014 at 7:53 am

    Jamie poses a fundamental question. The idea of running a flat pace is that it gives you the best time while making best use of fat-burning vs the scarce resource of glycogen stores that are used at higher paces. They can’t be replaced to keep up with the speed. IIRC you can keep going a very long time at somewhere around 72 to 75% of max HR, but the higher over that you go, the more energy comes from limited glycogen stores, not just linearly but with an ever-steeper curve. Of course, race tactics is a different matter.

  10. Chris Grauch on December 3, 2014 at 1:19 pm

    Great stuff Andrew!! Thanks for putting this out. I’ve often wondered about these exact things.

  11. Pete K on December 4, 2014 at 9:29 am

    Interesting read. Thanks for putting this together! It would be neat to see if this holds true for a larger sample size, ie. everyone in the elite field or all of the 50M finishers.

    Good luck to all of you who are running this weekend! I’ll be following via iRunfar.

    • Andrew Skurka on December 4, 2014 at 9:34 am

      Take a look at the Western States analysis that was linked in another comment. It actually shows that there is even more variance in the back of the pack, i.e. they slow down even more than those at the front, and increasingly so, not proportionally. I recall pacing my wife in the 2012 TNF marathon and my observations were consistent with these findings: runners who had no business running some of the climbs in the early miles of the race tried running them, and they paid for it dearly towards the end.

  12. Mike on December 5, 2014 at 8:41 pm

    Strava is not a good source for accurate info.

    • Andrew Skurka on December 6, 2014 at 4:54 pm

      I’m literally sitting in the same room as someone who works for Strava. She says that, indeed, the data on Strava can be inaccurate, but it’s a function of the quality of the data that is gathered by the GPS devices being used; it has nothing to do with Strava. The accuracy of GPS data is adversely impacted by the satellite connectivity, ping/breadcrumb interval, and (if you are using the Strava app) cell signal connectivity.

      Despite these limitations, Strava is the best there is.

  13. […] and I started running and I decided to take it out rather conservatively, because I had read an article about how too many people run hard in the beginning and can’t sustain their pace towards the end. […]

  14. Andrew Skurka Interview on September 25, 2015 at 7:27 am

    […] This is the type of analysis Andrew did for the TNF field last year. […]

  15. Adrian on October 1, 2015 at 3:42 pm

    Very nice article. I’m not quite in your league and I use different methodology, But I do spend a lot of pre-race time coming up with pacing charts. Personally, I try to forget about the pacing chart once I’m in the race and just go by feel, but I find it an absolute necessity for planning drop bag gear and narrowing time windows for family, crews, and pacers. Here’s what I did for Leadville –
    I also couldn’t agree with you more on runners starting out way too fast. It’s funny how every good marathon runner tries to run negative splits, yet when it comes to ultras, that seems to go out the window. Granted, I haven’t run negative splits on a 100 (yet), but my best 50 miler and 100K were evenly split. My theory is always “if you wouldn’t be running an uphill in the last 10 miles of a race, you probably shouldn’t be running it in the first 10 miles”.

  16. […] and gear, but he also backs up his opinions with research and data. Love it. One of his articles, Pace charts for TNF 50. And how everyone starts too fast, seemed like a good one to put to the test, using my gps data from the recon […]

  17. Almostran on March 10, 2017 at 5:53 am

    TY for sharing well done research. I think I achieve a faster run if I pay attention to how my efforts feeling than watching the data. This makes my runs more like a fartlek than an evenly paced run.

Leave a Comment