This is part of a multi-post tutorial on creating a pace chart for an ultra marathon. Start with the Introduction, or skip to a specific section:
Before I jump into the third and final step in creating a pace chart for an ultra marathon, I thought it would be helpful to review the process so far.
Step 1: In a spreadsheet we created a list of key landmarks on the course, and added pertinent information like distances between these landmarks and where drop bags can be sent.
Step 2: We established a goal time for the race, using training data, past race performances, or the ITRA index.
In this final installment, I will explain how to:
- Find splits between course landmarks, and then
- Correlate these splits to your goal time.
Let’s get started.
Finding splits for ultra marathons
Several ultra marathons have frequent timing stations (e.g. at every aid station), with race officials or automated timing systems recording the cumulative time and placement for each runner. A subset of these races will share the data, probably on their websites.
One example is Western States 100, which publishes splits from about twenty landmarks along the course. The page has a disclaimer (“unofficial times [that] may not be completely accurate”) but in this particular case I’d assume that the splits are generally trustworthy.
Data mining on Strava
If official/unofficial split times are unavailable, the backup option is to use Strava. More likely than not, this is what you will have to do. A few comments:
- You need to find runners on Strava who ran the race previously (including yourself) and who uploaded their race data. Many ultra runners are on this platform and normally you can find a few for any moderately popular race that’s been around for a few years.
- If you have your pick of the litter, select runners who executed smart races (i.e. conservative start, steady effort, relatively strong finish), regardless of whether they finished near your goal time or not — we can account for that later.
- Strava’s search tool allows you to search athletes and activities. The athlete search is a challenge for common names like “Nate Jones.” In those cases, try finding their Strava profile through their blog, or contact them directly for a link. The activity search will produce results for everyone, including lots of runners whose racing style you would not want to emulate.
- In longer races like 100-milers, GPS watches are vulnerable to dying unless measures are taken to extend the battery life of the watch, like changing the GPS interval to 5 seconds instead of 1 second. If you must, you can use data from a runner whose watch died 12 hours into a 20-hour race, but it’s best to find complete data profiles.
Using the topographic map and the time plot on the Strava activity page, manually record the splits for your selected runners. I select the first split available that is after my chosen landmark, e.g. aid station. The data available on Strava is insufficient to get an in/out time at aid stations.
Tip: As you scroll across the time chart, you can freeze your location with a left-click. This allows you to mouse over to another window/program without losing your place.
If you don’t know the location of the landmarks, you will have to find maps with their locations (or written descriptions). To start, check the race website. If the official maps are unsatisfactory, use CalTopo, which is a kick-ass mapping platform.
In some cases, there are telltale signs of the location of an aid station. For example, you might notice that the runner’s heart rate dropped significantly for a minute or two, or you might see some random directions in their GPS track (when they might have been searching for their drop bag or when they jumped into a portable toilet).
Take the average of multiple runners
After recording the splits for your chosen runners, average their splits. This helps to negate individual experiences or strengths/weaknesses that you are unlikely to share. For example, a runner:
- Stops for 5 minutes to empty their GI in a portable toilet;
- Hauls on the downhills, but is a relatively weak climber; or,
- Slows down while enduring a violent hour-long thunderstorm.
Correlating splits to a goal time
After averaging the splits of three runners, suppose that their averaging finishing time is 19:17. But your goal time is about an hour slower, in 20:15. How do you reconcile the difference?
The final step in creating a pace chart is to adjust the splits to reflect your goal time. I’ve had good success in doing it on a proportional basis. In the example above, a 20:15 finishing time is 5.0 percent slower than 19:17.
To create splits that will lead me to a 20:15 finish, I simply add 5.0 percent to the average splits of my chosen runners. For example, if on average they reached the first aid station in 40 minutes, I should plan to get there in 42 minutes (because 40 * 1.05 = 42.0).
Last year I ran the San Juan Solstice, a fantastic 50-miler in southwest Colorado.
Here’s my list of landmarks, from Step 1:
I established a goal time of 9:15, based on a past result (from 2008) and on comparative results with Jason Schlarb — we both had competed in Run Rabbit Run the previous year.
Next, I found Strava data for Schlarb, Brendan Trimboli, Michael Barlow, and Ryan Smith. Of these four, Schlarb and Trimboli ran the smartest races, so I took an average of their splits. Barlow and Smith had solid finishing times but faded badly throughout the race. I didn’t use their data because I don’t like to race that way.
When I took the average of Schlarb and Trimboli, the finishing time was 8:28. My goal time of 9:15 was 9.2 percent slower. So I added 9.2 percent to all of their splits, to create my own pace chart for a 9:15 finish. Here it is:
You can search for bighorn by going to the Strava search icon, select activity instead of athlete, and add city state: big horn, WY and keywords: Bighorn 100.
Ah, brilliant. Why had I not noticed that before?
Does it work well for you? When I search for an activity using Dayton, WY and “bighorn 100” no results come back.
It works great for me, there are about 4 pages of results. Did you make sure to select “Run” for the activity ? “Ride” seems to be the one selected by default.
Seems to be working for me now. Maybe the page needed to be reset.
Also, you can use Suunto Movescount to find activities and routes. There are not as much available than on Strava, but it’s a good source as well :
Another great source I typically use is Garmin Connect. It has a LOT of routes and I am usually sure to find what I look for 🙂
On Movescount you can use the Heat Map to find routes, http://www.movescount.com/map, but I have not figured out how to pull splits from the data.
As a non-Garmin user, Connect seems locked out to me.
For the activities found on Movescount, I usually export them in GPX and open them with a training analysis software to have access to the splits (SportTracks, PyTrainer, GoldenCheetah….etc) (I have a full list here if you need : http://www.freemovin.com/index.php/tracking-and-analyzing-part-1/)
As for Garmin Connect, I don’t own a Garmin myself, I just created a free account and use their “Explore” feature from there.
Great read! You summarize the method really nicely!
I used this method to calculate the pace chart for our run next month at the Eiger Ultra Trail. However, I did not only average over some runners, but calculated averages for elite, mid-pack and back-of-the-pack runners. There were interesting findings: Elite runners tend to start too fast (really much too fast), but back-of-the-pack runners start way faster (normalized to their average speed). Relatively speeking, the slower a runner is, the more the difference between his speed in the early hours and the speed close to the finish line.
So: If you use the splits as “role model”, you are trapped in the group-think, and this group-think is not the smartest way to run a race. On the other hand: In bigger races it is really hard to escape the hustle in the early kilometers.
For calculating my pace chart at races with a hilly or rolling profile, I also use a method based on some old examinations of McMahon (Muscles, Reflexes, and Locomotion. Princeton, NJ: Princeton University Press, 1984). He measured the power/exertion for running/walking at several inclines, and calculated a kind of calibration factors for each incline. If you know the distance of the race or race segment as well as the meters of ascend and descend, you can calculate with these calibration factors, to what distance this would match (if the power/excertion would be the same), if this distance was perfectly flat.
From there it is easy to calculate a pace chart which helps you to always run at the same level of excertion regardless of the actual incline. (This model, however, has also some limits, e.g. it does not take into account for nature or technicality of the route).
I have written two blogposts about this method – however: they are in German (don’t know, whether you speak that language).
Correct: Everyone starts off too fast relative to their eventual average pace, but back-of-the-pack runners slow down much more dramatically than the leaders.
The winners are winning, but I agree that they may not be doing it in the most optimal fashion. Can you run an even (or negative) split in an ultra? Gee, I don’t know. I suppose you could, but the ultimate limitation might be your patience. If you want to win the race, are you really going to run with back-of-the-packers for the first few miles, middle-of-the-packers for the next 25, front-of-the-pack through 75, and then catch all the leaders in the final quarter?
What you call “group-think” I describe as “peer pressure.” Early in a race, no one wants to be the first runner to start hiking. And no one wants to bottleneck up the rest of the field behind them. That’s why I like to start further back, so that I can run very comfortably among less talented runners who are starting off at a relatively higher effort.
You are right: Given all these pressures – from yourselfe (lack of patience) and from others (peer pressure … that is a good term for it) it is really hard to run a constant pace. And in many races you have indeed some street or fire road for the first mile, which then leads to a single track. And everyone in the pack is fighting the bottle neck … although they might create just that going an unnatural high pace.
Good luck for your races – saw that you will race UTMB. Nice trails out there! (although unfortunately the fight over “payed qualification points” at the moment is not nice at all).
I enjoyed following along and working the numbers for a future race. Thanks Andrew!
Thanks again for this guide, Andrew. I used it a few months ago for a 50k and just used it for my first 100, Headlands. I appreciate it!