About the Author
Albertus Roux is a sports scientist and ultra-trail runner specialising in endurance performance and human physiology. Drawing on scientific research, coaching experience, and years of ultra-distance mountain running, his work focuses on resilience, discipline, and the mental and physical demands of endurance sport. He helps athletes improve performance and unlock long-term potential through evidence-based insight and real-world experience.
By Albertus Roux
Part 1 of a 3-part series based on my Honours dissertation: A multivariate model predicting ultra-trail running performance.
Ultra-endurance sport has grown rapidly over the past decade, with events like Ironman and the Comrades Ultramarathon leading the way. Ultra-trail running, as a subcategory, has expanded into a global sport with dedicated world series events, professional athletes, and runners of all abilities pushing their limits.
Nowhere is this growth more visible than in South Africa. With our unique terrain, challenging courses, and strong endurance culture, trail running has become a natural extension of the country’s sporting identity.
As participation increases, so does the need to better understand what actually drives performance. Research in this space informs how we train, fuel, recover, choose equipment, and coach athletes. But ultra-trail running remains notoriously difficult to predict due to its complexity, environmental variability, and sheer duration.
That’s exactly why it’s worth studying.
Why Try to Predict Ultra-Trail Running Performance?
The goal isn’t just to predict race times. It’s to explain performance.
When we understand why certain athletes perform better than others, we can train with direction instead of guesswork. Ultra-trail running requires a massive investment of time and energy. Identifying patterns linked to strong performance helps ensure that this investment is spent wisely.
In a sport filled with uncertainty, these patterns give us something solid to lean on. They highlight principles that tend to hold true across different races, conditions, and athletes.
In research, these patterns are referred to as variables.
What Are Performance Variables?
Variables are measurable, numerical factors that can be analysed statistically. Examples include training volume, elevation gain, personal best times, physiological markers, and race-specific metrics.
In quantitative research, like this dissertation, variables allow us to move beyond opinion and intuition. They give us a structured way to examine what actually matters for performance.
But identifying variables is only the first step.
What Is a Correlation (and Why Does It Matter)?
A correlation describes the relationship between two variables. In simple terms, it tells us what happens to one variable when another changes.
For example, the faster your running pace, the shorter the time you can sustain it. These relationships are expressed using an “R value.” The closer the value is to zero, the weaker the relationship. The closer it is to one, the stronger the relationship.
Correlations help us identify which variables are most strongly associated with race performance. Once these relationships are identified, they can be explored further within a model.
What Is a Performance Model?
A performance model combines multiple variables to better explain performance.
Instead of looking at training hours or fitness in isolation, a model examines how combinations of variables relate to race time. For example, how training volume, elevation gain, and a 5 km personal best might interact to influence ultra-trail performance.
In some cases, these models can even provide rough performance predictions based on known athlete data. More importantly, they help us understand which factors are worth prioritising in training and coaching.
The Classic Endurance Model (and Why It Falls Short)
The classic endurance model consists of three physiological variables:
- Maximal oxygen uptake (VO2max)
- Lactate threshold
- Running economy
These measures are commonly used to compare endurance athletes and are highly effective in road running. In fact, this model has been shown to explain up to 94 percent of performance variation in 16 km road races.
For road runners, this model provides clear direction. Improve these variables, and performance improves.
However, when applied to trail running, the picture changes.
Research shows that the classic endurance model explains only about 50 percent of trail running performance over similar distances. In other words, half of what determines trail performance is not captured by traditional physiological testing.
As race distance and complexity increase, the importance of these classic variables appears to decrease even further.
This gap highlights the need to look beyond physiology alone.
Why Trail Running Is Different from Road Running
Trail running differs from road running in several critical ways:
- Larger elevation gains and losses
- Technical and uneven terrain
- Longer race durations
- Greater environmental stress
Running uphill and downhill over long distances causes significant muscle damage due to repeated mechanical loading. Technical terrain requires shorter stride lengths, faster reactions, and constant adjustments.
Trail races also last longer than road races of the same distance, amplifying fatigue, muscle breakdown, nutritional challenges, and mental strain.
All of these factors introduce demands that are not well captured by traditional lab-based measures.
Towards a Better Model of Ultra-Trail Performance
The remaining unexplained portion of trail running performance likely lies in how well athletes manage these unique demands.
Attributes such as fatigue resistance, local muscular endurance, confidence on technical terrain, and tolerance to discomfort are difficult to measure directly, especially over long periods. However, they tend to develop naturally through consistent trail-specific training.
This is where training characteristics and real-world performance tests become valuable. Metrics like training volume, elevation gain, and prior race performances act as practical proxies for these harder-to-measure attributes.
That said, traditional physiological measurements still matter. They continue to explain part of performance and should not be ignored. The goal is not to replace the classic endurance model, but to expand it.
By combining physiological data with training and performance variables, we can build a more complete picture of ultra-trail running performance.
In the next parts of this series, we’ll explore what this expanded model revealed in a group of 19 athletes competing in the 2024 Ultra Trail Cape Town 100 km.
FAQ
Q1: What determines performance in ultra-trail running?
Ultra-trail running performance is influenced by a combination of physiological factors, training characteristics, terrain management, fatigue resistance, and race-specific demands like elevation and duration.
Q2: Why can’t traditional endurance models fully explain trail running performance?
Classic endurance models focus on VO2max, lactate threshold, and running economy, which work well for road running but explain only about half of trail running performance due to terrain, muscle damage, and technical demands.
Q3: What variables matter most in ultra-trail performance?
Key variables include training volume, elevation gain, prior race performances, downhill resilience, and adaptability to fatigue, alongside traditional lab-based physiological measures.
Q4: How is trail running different from road running performance-wise?
Trail running involves greater elevation changes, technical terrain, longer duration, and higher muscle damage, making pacing, strength, and fatigue management more important than steady aerobic output alone.
Q5: Can ultra-trail performance be predicted accurately?
While no model can perfectly predict performance, combining multiple variables into a multivariate model helps explain performance patterns and guide more effective training and coaching decisions.
References
EHRSTRÖM, S., TARTARUGA, M. P., EASTHOPE, C. S., BRISSWALTER, J., MORIN, J.-B. & VERCRUYSSEN, F. 2018. Short Trail Running Race: Beyond the Classic Model for Endurance Running Performance. Medicine and science in sports and exercise, 50, 580-588.
JOYNER, M. J. 1991. Modeling: optimal marathon performance on the basis of physiological factors. Journal of applied physiology (1985), 70, 683-687.
MCLAUGHLIN, J. E., HOWLEY, E. T., BASSETT, D. R., THOMPSON, D. L. & FITZHUGH, E. C. 2010. Test of the classic model for predicting endurance running performance. Medicine and science in sports and exercise, 42, 991-997.
SABATER-PASTOR, F., TOMAZIN, K., MILLET, G. P., VERNEY, J., FéASSON, L. & MILLET, G. Y. 2023. VO2max and Velocity at VO2max Play a Role in Ultradistance Trail-Running Performance. International journal of sports physiology and performance, 18, 300-305.