Trail running fans are built different. A disturbingly dedicated soccer fan might wake up at 4 A.M. to watch a game of a 2nd-division English club team. The same trail running fan will not sleep at all to watch some dots move disturbingly slowly across a low-resolution map of the French Alps. WE ARE NOT THE SAME.
If you do enough dot-watching on the race live-trackers, you might start to notice a pattern: some time splits seem to be way more important than others. If an athlete moves up through the field while dropping down into Courmayeur at UTMB, or to Rucky Chucky at Western States, it seems like they often proceed to blow apart the front of the race. Meanwhile, other splits just seem like nothingburgers — lacking much drama or insight at all. Here, the word “drama” should be taken with a grain of salt for our average reader, since it’s coming from a writer who routinely pleads with God to make one of those dots get to the next split 5 minutes faster. The Creator acts via Livetrail, where He clicketh refresh.
Superfans of the sport may have a general feeling that some splits matter more than others, but is that supported by the data? A fantastic November 2022 study in the Journal of Functional Morphology and Kinesiology sought to answer that question, looking at 5 years of split data at four races to determine the terrain and pacing attributes that have the strongest associations with performance. It’s like Moneyball for trail running splits, going from gut feelings to hard data that could inform training and pacing strategies.
The study design would titillate any avid dot-watcher, compiling split time data for 5 editions of UTMB and CCC (2015-2019), and 3 editions of the Javelina 100 Mile and 100k (2017-2019). In total, 16,518 athletes finished the races and were included in data analysis, with incomplete datasets (DNFs and timing errors) removed. The trail between each checkpoint was classified as uphill, downhill, flat, or mixed, with the splits that were flat and mixed discarded. The remaining splits were divided into two categories: first half of the race, and second half of the race.
To compare across splits, average race speed was divided by average split speed to determine relative split speed. Data analysis included a couple subgroups: top 50% of finishers and bottom 50% of finishers, plus top-10 men and top-10 women.
My fellow dot-watchers probably just soiled their pants in excitement. We see all of these splits, and there seem to be patterns – but are there? Now is the time to take a step back and evaluate your own assumptions. This study design should give us some clues about uphills versus downhills in the first half versus second half of ultra races, for male and female athletes depending on their finishing place. What attributes are associated with athletes who excel?
The tempting answer would be to say that the fastest athletes crush the uphills, because that is where the aerobic and muscular demands are the highest. (Come on, I know you’re tempted to say that.) But the assumption that uphills are most important would be very wrong.
Let’s start with the no-shit findings. The relative split speed was lower for uphill sections than downhill sections by 57%. That makes sense, because uphills are designed by the Devil to test our faith (the Devil makes uphills on his lunch break from designing shoes with narrow toe-boxes). In addition, the relative split speed was 33% slower in the second half of these races across all splits. Ultras are impossible, so that result tracks with expectations too. The top 50% of finishers faded just 1.7% less than the bottom 50% of finishers, showing that almost everyone fatigues at the end of races.
Now we’re getting to the sexy findings. The relative split speed decrease in the second half of races was 24.5% larger for downhills than uphills. Everyone slows down, but an extra full diaper’s worth of that fade happens on downhills! That trend is 3.1% larger for the bottom 50% of finishers, meaning accentuated downhill fades may partially explain worse finishes. And faster finishers go quicker on downhills relative to uphills than slower finishers, with a relative difference of 2.6% between the groups.
I hope you saved some room in your colon, because the findings for elite athletes may be the coolest of all! In the second half of races, the relative split speed of elite men slowed down 17.9% and elite women slowed down 22.3%, with the decrease higher for downhill sections (15.9% higher for men and 20% higher for women). In a finding that was counterintuitive for me, elite women paced less evenly than elite men, fading more in the second half of races. But the big headliner finding applies across all finishers, from the front to the back of the pack: athletes faded significantly more on downhills than on uphills, and the extent of that downhill fade was the strongest predictor of finish time.
The study has plenty of limitations. First, tossing out the flat and mixed-terrain splits could obscure signals that drive performance. The same goes for DNFs. Second, the authors didn’t control for the average gradient. Given that UTMB’s 30% grades are lumped in with Javelina’s 3% grades, it could be a case of comparing apples and oranges and calling them “fruit,” which I’d be OK with, or comparing apples and Fritos and calling them “food,” which would be more concerning. We’d need more data to be sure how grade and race terrain drive the findings. Third, and most significantly to me, I’m curious about how the higher denominator in downhill splits may affect the results. At the start of a race, an athlete might go 8 miles per hour on downhills and 4 miles per hour on uphills, fading to 6 miles per hour on downhills as their legs fatigue, while still hiking 3.5 miles per hour on ups. If you start with slow pacing, like on uphills in an ultra, there seems to be less room (or physiological necessity) to slow down.
However, have faith! The general findings are echoed across other research. A 2014 study in the Journal of Quantitative Analysis in Sports looked at results in 7 different sub-ultra fell races in the UK and found that “faster finishers tend to be those with better descending skills.” A 2019 study in Frontiers of Physiology found that performance across individuals in a 7K, steep trail race varied most on downhills, with 25% performance differences on downs compared to just 10% on flats. In shorter races, downhill ability plays an outsized role on race results.
But what about ultras?
The big leap in the current study is to demonstrate through thousands of data points that downhill running variation and importance only get accentuated in ultras. That makes sense based on the physiology. A 2020 review article in Sports Medicine described how downhill running can produce mechanical breakdown causing central and peripheral fatigue. A 2007 study in the Journal of Sports Sciences showed just how much that can impact performance, with a single 30-minute downhill running bout at 15% grade causing a 7% increase in oxygen demand during subsequent level running, and up to a 21% reduction in maximum strength of the knee extensors. Those impacts likely accumulate over the course of a race, causing everything to get slower as the downhills mount up.
All of that data is cool, but what does it say for training? I think we can apply a few training guidelines based on the analysis of ultra splits, combined with the background of physiology and training theory.
One: Uphill speed matters, but think more about the starting point of uphill fitness than uphill endurance.
Even though uphill paces slowed down less than downhills, the starting point still mattered. A 2019 study in the European Journal of Applied Physiology found that uphill, level, and flat running economy are strongly correlated, and other studies show that uphill running has a higher demand on aerobic variables correlated with speed improvement. Long-term growth comes from aerobic and speed development over time, and uphill running ability is the better proxy for that growth. Wait, maybe the Devil designed uphills to make us faster?
Two: Downhill proficiency is a skill that needs to be continually developed.
Basically every study evaluating downhill running performance shows that it varies immensely even among athletes with similar fitness variables, especially on technical terrain. Since the aerobic demands of downhills are lower, the more heavily biomechanical and neuromuscular skill of downhills creates an opportunity for “free” speed for every athlete, at least from a metabolic perspective when compared to uphills. Practice faster downhills with good form whenever you have the opportunity, and especially within 8-12 weeks of races (5 training tips at this link).
Three: Downhill endurance is one of the most key elements of ultra performance, harnessed using the repeated bout effect.
The big breakthrough from the study is to reinforce the idea that biomechanical breakdown and neuromuscular fatigue from downhills doesn’t just cause delayed onset soreness in future days – it starts to torpedo performance as fatigue increases even within one-day events. Give me an athlete who can go uphills fast, and I’ll show you an athlete with a very high ceiling. Give me an athlete who can go downhills sustainably across many hours, and I’ll show you an athlete who can perform at the highest percentage of their ceiling. And that fatigue resistance is what makes ultra champions.
In the 4 to 8 weeks before races, focus on accumulating vert at ratios that are equal to or greater than the race profile. The starting point of speed/running economy measured via uphill running proficiency really matters when you get to that specific period, and it can improve across training cycles and raise the ceiling of performance. But when an athlete lines up at one of these ultras, the ability to withstand the downhills and reduce performance deterioration on them is what allows that fitness to shine.
I think that downhill ability is heavily influenced by internal narratives. Yes, we probably all have some baseline proficiency, particularly on technical trails, influenced by our training background and psychology. But from that starting point, the ensuing trajectory is the consequence of a series of daily decisions.
“I am not good at downhills” equals an athlete that gets more hesitant over time, adapting to that meekness at the cellular and systems levels. What starts as negative self-talk becomes crappy downhill physiology over the course of many adaptation cycles, a self-fulfilling prophecy that seems to act as proof for itself.
“I am a downhill beast” equals an athlete that approaches each downhill opportunity with a small sliver of increased confidence. Because downhills are less metabolically limited, “fitness” is rarely what’s preventing athletes from going faster, so each ounce of confidence leads to slightly faster paces and fewer fades. What might start as a second per mile difference balloons into a minute per mile, psychology driving physiology through adaptation to the new training stimuli.
Let’s be real: no matter how much we butter our own downhill running bread, none of us will be Kilian Jornet, glissading down mountains as if our 23andMe says we are 12.5% mountain goat. But by thinking a bit more like Kilian, and training accordingly, we might find a great reserve of potential that has been hiding out all these years.
David Roche partners with runners of all abilities through his coaching service, Some Work, All Play. With Megan Roche, M.D., he hosts the Some Work, All Play podcast on running (and other things), and they answer training questions in a bonus podcast and newsletter on their Patreon page starting at $5 a month.