Are MTU™s an adequate measurement to compare mountain and road races?

Comparing relative effort is something that has been done in the running world for many years, using the World Masters Athletics age and sex grading process. Simply put, age and sex grading compare your race time to the world record for your specific age group and sex, allowing for comparison between performances of runners of different sexes or age groups. For example, a 75 year old male who runs a 45 minute 10km would be equivalent to the same effort of a 25 year old running 10km in 31 minutes. Learn more here

Using the underlying principles of age and sex grading, we can use a similar method to compare mountain and flat courses. To make this comparison, we are assuming that a course record is essentially the World Record for that particular race course. 

By plotting the World Record time at each commonly raced distance, you can explore the relationship between the two variables and end up with a very strong correlation between them. When analyzing Men’s world records, this results in an R2 of 0.999 for races from 800m to the Marathon (Fig. 1), and R2 of 0.998 for races from the marathon up to 24 hours (Fig. 2). The same trend is seen with Woman’s world records with R2 of 0.999 and 0.993 (Fig. 3, and Fig. 4)

For this comparison of road, track and mountain races, 22 world class marathons and ultramarathons between 42.2km and 303.51km were selected to analyze, and Men’s and Women’s course records were used. There are many highly competitive races out there, but for this comparison, we selected races that are well known with a highly competitive field, and the results were easy to find and thus compile. These races include:

After plotting Men’s and Women’s course records for each of these races (Fig. 5 and Fig. 6) you can see there is quite a bit of variability, with the R2 dropping to 0.805 for men’s races, and R2 of 0.777 for women’s races.

The largest variation in course records at a single distance can be seen at the 100 mile distance, for both men and women with course record times ranging from the world record of 11:19:13 (on a track), all the way to 23:28:10 (Hardrock counterclockwise) – more than double the time for the exact same race distance.

After adjusting the data, and plotting race time vs MTU™ rather than distance (Km), we see this variability dramatically decrease (Fig. 7 and Fig. 8), as the true effort required to run these courses is revealed. For instance, races with massive elevation gain like the Hardrock 100 (10,074m of climbing) suddenly become 261 MTU™, while the 100 mile world record run on a flat track stays at 161 MTU™. Adjusting Km to MTU™ brings the correlation of these races to 0.977 for Men, and 0.965 for Women, revealing that MTU™s may do a better job of comparing effort when individual course factors like elevation gain vary.

But 0.977 still isn’t the near perfect correlation of 0.999 we see in the WR comparison.

This difference, we believe, can largely be explained by sample size.

With road and track racing, courses are much more similar, and each distance is raced far more often. For instance, in the USA alone, there were 16,500 5k races 2015, with over 7.6 million finishers. By comparison, even The Ultra-Trail du Mont-Blanc (UTMB), one of the largest trail ultramarathons, limits the field to 2300 participants every year, meaning that there is a much smaller sample size of performances (2300 compared to 7.6million+ per year).  It is likely that the course records  for these 20 races have not yet been tested to the same limits, and as a result we are likely to see improvements upon these times in the years to come.

As you may notice, the Hardrock 100 falls far outside the line of best fit. It is a good example of this sample size effect. Hardrock is limited to 140 racers per year, chosen by a very selective lottery draw. What is more, the Hardrock 100 also alternates direction every year. As a result, both Hardrock 100 courses (clockwise and counterclockwise) have only been raced 12 or 13 times each by a relatively small number of people. As a result, the two course records for Hardrock lie considerably above the line of best fit, meaning they are slower than you would expect for the given number of MTU™s (Fig 9).

Finally, there is the effect of terrain. No measurement (MTU™, GAP, TSS) has been able to account for this yet, and I don’t know if anything ever will be able to. Obviously running in ankle deep mud, or picking your way up some class 5 scrambling terrain is slow going, and the lack of speed isn’t caused just by the horizontal distance or elevation gain. Terrain matters, so if you know a course or conditions are much more technical than the norm, take the MTU calculation with a grain of salt (that might even help you stay hydrated while you are running).