If you’re a fan of professional cycling, and have followed the sport partly here on the world wide web, you have likely been exposed to some level of controversy around the so called “suspicious, or mutant cycling performances”. Just like allergies, this seems to be a recurring phenomenon blooming every year, beginning with the occurrence of major Grand Tours.
Recent years have seemed worse though. This could be related to an increasing understanding that the grand public has of cycling power related data. Power meters facilitate quantifying performance. It makes things easily comparable.
At Trimes, given the complexity of most elements involved in this case, we thought it would be best to kindly invite one of the leading figures of cycling power data models.
Known for having set several standards in this regard, beginning with Training Stress Score (TSS) in early 2003, to its refined evolution into the Performance Management Chart (PMC) in 2006, Dr. Andrew Coggan, Ph.D in exercise physiology (Fellow of the American College of Sports Medicine), has from day one been known to be ‘annoyingly’ obsessed with accuracy. After all, if it hadn’t been the case, we wouldn’t trust his works. He is the co-author of “Training and Racing with a Power Meter”, handling the scientific component of this best selling book.
Dr.Coggan, or just Andy, as he likes to be called, has also recently been known for studying cycling aerodynamic drag, or CdA, that is, the air resistance attributed to a cyclist. These works see him moving from indoor wind tunnel testing to outdoor field testing. In short, he is after finding ways to test outdoor, one’s CdA, as it turns out to be closer to the reality, i.e. more specific.
Therefore at Trimes we thought that the combination of his deep knowledge in exercise physiology, cycling power models and a strong interest in how wind (and other resistance inducing parameters) affects performance, made Dr. Coggan a pragmatic choice for this interview.
We will stick to a few very basic questions, and more importantly we will avoid mentioning names or referring to any particular article.
Dr. Coggan, first thank you so much for having accepted this interview.
1. Do you think that your works along with available analysis tools, such as the excellent works of Tom Compton’s Analytic Cycling website, can be used to to estimate the power that a rider generates over any given intense segment of say, an uphill race? Within a reasonable error margin?
The physics of cycling are rather simple and well-understood, so it is quite easy to derive an *estimate* of a rider’s power output from a given set of inputs (e.g., mass, aerodynamic drag characteristics (i.e., CdA), rolling resistance, etc.). The difficultly lies in obtaining sufficiently-precise values for such inputs. For example, even on a moderately-steep climb speeds are often high enough that small differences in assumed CdA and/or in actual (but unmeasured) wind speed can have a significant impact on the final value of power output obtained. Moreover, such errors/uncertainties aren’t just additive but multiplicative in nature. Although I am a research scientist who only dabbles in the applied end of things as a hobby, I am fortunate to see many actual power meter files from professional cyclists, including some from riders fighting for placement on the GC in grand tours. Based on that experience, it isn’t uncommon to see a difference of ~5% between the estimated and actual power output, with some outliers differing by up to ~10% (as any power meter-using cyclist who has ever uploaded their data to Strava can attest).
2. Thank you. Your explanation seems sound to me. However a 5% error margin is still a bit above the results that some claim being achievable. I’m not questioning your point of view. You’ve been working with power since 1980. But just to be 100% sure. For let’s say, 1 Million dollars! Could you work a bit harder and try to bring this error margin straight under, say, 2%?
To some extent, I think I addressed the answer to your question (how accurate can such estimates be?) to at least some extent in response to your 1st question. That is, it is mostly a case of GIGO (garbage in, garbage out) rather than any fundamental problem or discrepancies between people in how you get from climbing speed to power. IOW, « for a million dollars » you could undoubtedly come up with rather accurate estimates, e.g., by placing wind sensors every 100 m along a climb, by paying the riders to undergo wind tunnel testing so you knew their true CdA, by precisely weighing their equipment, etc. Unfortunately, in the vast majority of the cases we don’t have access to such data, but instead must rely on guesses and inferences, which add imprecision.
3. Point taken Dr. Coggan. At least I tried, but I don’t own a million dollars anyway.
So I’m concluding that speed to power estimates may not be the best way build a doping suspicion case against a rider. To the best of your knowledge, would there be a better solution? We’ve heard of a method called VAM.
VAM was a term first popularized by a famous Italian trainer, and is the acronym for “ velocità ascensionale media” which basically translates as “average ascent speed”. Would *this* be reliable? Could this allow for accurately comparing performances across riders?
First, back to question #2 a bit: I actually don’t think there is a large difference of opinion re. the accuracy of estimated power outputs. That is, I think everyone would agree that, *on average* they are (or at least can be, if you know what you’re doing, and have access to the requisite input data) good to w/in a few percent. The problem, though, is that is only *on average*, whereas what matters when applying such calculations to any given individual is the accuracy in that particular case. Based on the powermeter files I have seen from riders racing in, e.g., the TdF, it isn’t uncommon for estimates for individuals to be off by ~5%, and sometimes as much as ~10%. IMO, that is simply too much « slop » for such calculations to add anything of merit to any doping debate.
As for VAM, it is better in the sense that it relies on far fewer assumptions to calculate. OTOH, it is still dependent upon the exact conditions under which it is measured, e.g., slope, length, and placement (both within a stage and within the stage race as a whole) of a climb, wind, how things play out tactically, etc. Thus, I think it is best to just compare apples-to-apples, i.e., to only compare performances on the same climb (and even then you have to keep in mind the limitations mentioned above, at least if comparing across years).
4. Let us “bunny hop” over the ethical aspect of this and assume that we could indeed have access to World Tour riders’ power data. Let’s also assume that this data is gathered using well calibrated units, and that the files were not deliberately manipulated.
Could abnormal performances be detected by inspecting an event’s power file? I guess that in order to achieve this, one needs to draw a doping power plausibility line (expression borrowed from Alex Simmon’s blog)? Could we, or science define such a line?
In a word, no.
An individual’s sustainable power output is a function of the maximal oxygen consumption (VO2max), the fraction of their VO2max at which they can exercise for a prolonged period of time, and their mechanical efficiency (i.e., ability to convert metabolic energy into useful external work). If you take the highest published values for each of these three from the literature and then assume they occurred in the same individual, you can calculate that the upper limit to sustainable human power is >9 (!) W/kg. Even if believe some of these reported values are in error (and they very well could be) and assume more “pedestrian” values (e.g., VO2max = 90 mL/min/kg, 85% of VO2max for 1 h, 24% gross efficiency), you can justify a power of 6.65 W/kg for 1 h as being “physiologically plausible”. While such high power outputs have never been previously reported (the highest, directly-measured 1 h value of which I am aware is just shy of the latter figure…although of course there is no way of truly knowing whether or not the individual was doping), this does not mean that they are impossible.
It simply demonstrates that such high values for VO2max, fractional VO2max, and gross efficiency have never occurred in one person (or if they did, the person took up, e.g., needle-pointing instead of cycling). Furthermore, there is no sound physiological reason why this could not occur…in fact, with the increased globalization over the last ~20 y of what used to be almost exclusively a western European sport, the odds of finding a “Secretariat” cyclist would seem to be higher than ever.
Indeed, a recent scientific review identified some 40-odd genetic variations associated with elite athletic performance, but noted than no one individual had ever been identified who possessed more than about 20 of them. So, when people start declaring that there is an obvious “doping line” that can’t be crossed, I just have to shake my head in wonder.
5. In pretty much every sport, humans have constantly improved, pushing the limits further, decade after decade. In a recent blog article, Australian cycling coach Alex Simmon from RST Sports LTD was outlining this applied to the world of swimming, by plotting the evolution of world records over the 1500 Freestyle, since the event has existed (see useful links section below).
Andy, do you think that it’s reasonable to believe that this natural evolution – resulting from the advances in technology, better training methods, probable genetic adaptation, etc – could allow nowadays (as well as future) *clean* riders to beat *doped* riders belonging to the past?
That’s a good question, and one to which I don’t think I (or anyone else, really) can provide an easy answer. Certainly, though, it would be rather illogical to believe that we’ve reached the limits of human performance in cycling only 150 y after the bicycle was invented, when it took 1-200,000 y to achieve the much longer-standing goal of man-powered flight. Thus, I would argue that, as in other sports, cycling performances will continue to improve (at a faster or slower rate), such that eventually non doped performances will surpass those of admitted dopers. However, when that might happen, if it hasn’t happened already, is really anybody’s guess, especially when you consider how records do not improve gradually/smoothly, but rather in fits-and-starts.
Dr. Andrew Coggan, thank you very much for your time. You’ve provided us with a good start. Readers eager to learn more on this topic, are kindly invited to visit Alex Simmon’s cycling blog, for he has posted a few compelling articles looking at this case from various angles.
By Charles G. Couturier, triathlon coach
With the help of:
Andrew Coggan, phd
Suzanne Atkinson, md
Andrew Coggan, on Training and Racing with a Power Meter/
Training Peaks’ Power 411 landing page
Alex Simmon’s blog
The evolution of the 1500m Freestyle
The Elusive Dopeometer
Cycling Power Lab