🚨Calling cycling Twitter: VAM ≠ watts/kg🚨
Since the start of 2020, climbing times and race speeds have taken followers of professional cycling by surprise. Times set by riders previously convicted (or suspected) of doping on famous mountains in big races have been beaten by allegedly clean riders. This has set off a cascade of people - generally on twitter - suggesting this new generation of riders are doping by citing their watts/kg calculated from Velocità Ascensionale Media (VAM - the average ascent speed of a rider).
It's my assertion that using VAM to calculate a watts/kg figure is not enough to accuse somebody of doping because of how error riddled the calculation can be. But first, I want to start off by accepting that I too am sceptical of some of these performances and respect fans who stand by the principle of curiosity. We all want a cleaner sport and we all want transparency. The more progress we get towards these goals, the better.
What I struggle to get on board with is when this scepticism turns to cynicism.
As a community, we will do better to keep an open mind rather than defaulting to conspiracy theories so I am open to hearing the best good-faith criticisms to this piece in the comments. As the political scientist Philip Tetlock suggests; our beliefs are hypothesis to be tested, not treasures to be guarded.
So let's kick this off. Why has there been broadly shared amazement at recent climbing performances?
Times set by riders such as Tadej Pogacar, Tao Geoghan-Hart, Simon Yates, Egan Bernal, Jai Hindley, and Mark Padun have all received the suspicion/cynicism treatment often after breaking records set by riders who were either definitely cheating, or at least likely to be cheating. VAM - an indicator of how fast riders are climbing - has traditionally been thought of as useful for comparing climbing efforts on different hills and has also been used to compare riders over generations. It’s also often used to estimate the watts/kg of a given rider. However, it's my point of contestation that VAM is not a good proxy for measuring watts/kg, and therefore it becomes incredibly difficult to compare across different climbs, or compare across different generations of riders.
Ultimately, VAM is measuring the relative speed it takes a rider to gain a certain amount of elevation, while watts are measuring the power output of a rider. While there is undoubtedly a correlation between the two, there is more to speed than just power.
So what can influence the speed of a rider climbing a mountain outside of pure power output?
Gradient: Despite VAM rising exponentially with gradient, climbs with shallower average gradients are going to give off disproportionately higher watts/kg scores for riders than climbs with steeper average gradients because drafting allows those in the slipstream to use less energy but retain the same speed. This gets even more complicated when climbs vary in their gradient between steep sections and flatter sections because the drag coefficient is not constant throughout the climb. You would expect that over longer climbs with lots of fluctuations in gradient, errors would be multiplied further, so VAM scores will likely become less and less accurate for calculating watts/kg. This also makes it difficult to compare climbs, even if they have similar average gradients - for example La Plagne (mostly steady) and Piancavallo (more variation in gradient).
Aerodynamics: Similarly to variations in gradient, riders who are more aerodynamic tend to go faster even on climbs. Enough research has been done now to suggest that air is still a significant drag on a rider's speed at 6% gradient. This means that on the parts of climbs where it's 6% and below, those with aero bikes, aero kit, and are drafting in the bunch will score high VAM scores with disproportionately low watts/kg. This is even more significant now with aerodynamic bikes that weigh close to the UCI limit.
Rolling resistance: Cannondale's research has suggested that at a 10% gradient, rolling resistance contributes to roughly 10.2% of the force the rider has to overcome. At lower gradients it is slightly more. This is because it is a function of velocity (when you are going faster, the wheel rotates more often, so rolling resistance has more of an impact). Still, reducing rolling resistance can be significant on climbs in the World Tour when races are frequently decided by seconds in an hour long race. So riders running faster tyres and more optimal tyre pressure will ride faster than those who don’t. This is a significant difference between generations of riders. Furthermore, climbs raced today are often resurfaced (such as the Passo Fedia in the 2022 Giro d’Italia) meaning comparisons to past generations become even more inaccurate. Faster riders means higher VAM, giving the impression of high watts/kg.
Atmospheric conditions: This has the potential to be one of the biggest impacts. High wind speeds can be incredibly favourable for riders' VAM while they may still only be hitting (relatively!) modest watts. Anybody who has taken a Strava KOM by surprise will know this all too well. While the impact of wind speed is obvious, measuring the impact it has during the race can be incredibly difficult. Real life speed and direction of wind often vary from forecast conditions which means that controlling for this variable between climbs and between generations makes VAM a poor way of calculating watts/kg.
Weight: Calculating an accurate rider and bike weight is essential in order to calculate an average watts/kg. However, even this is notoriously hard. Riders can lose up to 5kg just of fluid on the bike during some stages meaning, even if one had access to a morning weigh in of a rider (which is almost never the case anyway), estimating the total weight of the rider and their bike at the time of climbing will likely have a large room for error.
Racing a climb in a group, that varies in gradient, on an aero climbing bike, with fast tyres, optimal tyre pressure, and favourable wind conditions can feasibly give off a watts/kg score that seems implausible without doping. The point here is that controlling for the above measures between riders, between climbs, and between generations is really hard.
Furthermore, when a rider sets a "thermonuclear" time, twitter tends to react in predictable hysteria.
But virtually no attention is paid to the riders finishing seconds behind. If one logically deduces that a rider must be doping by using VAM to infer their watts/kg, then someone setting virtually the same watts/kg just a handful of seconds behind must also be cheating. For example, Wout Van Aert broke Chris Froome’s climbing record on the Prati di Tivo in 2021, 45 seconds behind Pogacar in 9th (18 other riders that day also broke Froome’s record of 38 minutes and 22 seconds). Similarly, Richie Porte broke Miguel Indurain’s record on La Plagne in the 2021 Dauphine when he finished 34 seconds behind Mark Padun on stage 7.
It’s my hypothesis that if everybody on a given race day is setting a fast time on the climb, then factors other than power are likely to be heavily influencing the time.
We all accept that VAM is used as an educated way of guessing watts/kg, but nobody seems to have questioned how educated these guesses are. While those who calculate these metrics accept a degree of uncertainty, figures are never presented with a lower and upper bound. Nor do we know how sure one is on a given calculation (say, 80% sure that rider ‘abc’ hit ‘xyz’ watts/kg). Instead, our critical thinking gets hijacked by our intuitive response when we see a high watts/kg score without stopping to question how it was calculated or how accurate those numbers actually are.
Let me say now that this is not a criticism of those who put in the time to try and calculate watts/kg from VAM (I am a huge fan of Mihai and ammattipyöräily’s twitter content and you should definitely follow them both if you are not already doing so). I acknowledge that VAM is currently the best way of making an educated guess. I also recognise that watts/kg is still the single biggest factor influencing VAM, especially on steep climbs. However, it is nevertheless important to acknowledge the limits of using VAM as a proxy for watts/kg and, ultimately, to take the estimated numbers we see on twitter with a pinch of salt.
Fundamentally, as the sport progresses technologically, riders are able to go faster and faster without having to necessarily hit proportionally stratospheric watts. But this gives the impression, through VAM, of riders hitting the same watts/kg of doped riders.
So VAM ≠ exact watts/kg. This means using roughly calculated watts/kg can’t be reliably used to accuse somebody of doping. But now for an essential part - what would change my mind?
Obviously, if there was a major (>10 riders or >2 teams) doping scandal then I would find it easy to quickly revaluate some of the performances mentioned above. However, I think if there was a large scale statistical analysis showing a consistent correlation between watts/kg and VAM, over different climbs with different riders in different races, I would revaluate my belief. This study would be really interesting to attempt but would probably need widespread cooperation from professional teams which is currently unlikely.
I think I would also change my mind if climbing times rise exponentially over the next handful of years. This is because I think a rate of increase that quick would be faster than what current advances in training, nutrition, equipment and aerodynamics would be able to account for. So if the Mount Ventoux or Alpe d’Huez records are broken by over 2 minutes in the next 3 years, expect me to be significantly more sceptical.
It’s worth bearing in mind, however, that today’s riders may actually be better than previous generations (for example, Bjarne Riis mentions in Daniel Friebe’s new book that riders of his generation were lazier in training and lifestyle because of the power of EPO). Since the Armstrong years, journalists and fans alike seem very wary not to be ‘caught out’ by believing in riders. But, as mentioned previously, defaulting to cynicism is hardly a viable policy for determining the truth. Progress in all areas of the sport (from increased grass roots competition, to advancements in nutrition) mean that it’s perfectly possible that more of today’s young talents are able to perform at a higher percentage of their potential, more often. This would explain the proliferation of fast climbing times, as well as the recent rise in talented youngsters. Whatever the reasons for this rise in speed in the professional peloton, we should at least be trying to offer testable explanations.
I do not know whether the top riders in the mens sport are doping or not. But I hope I have convinced you that using VAM data to calculate watts/kg, which is then subsequently used to accuse somebody of doping, is often riddled with caveats.
As mentioned at the start, all good-faith criticism in comments will be appreciated. It’s my worldview that, as a community, we should all be trying to form beliefs that are accurate, rather than believing things that only make us feel good. This means listening closely to ideas from the people we disagree with. I hope this post can be a step in the right direction.
Thanks for reading.