Cycling Science: Is power output the key to catching dopers?

The 2015 Tour de France is over and with it the endless stream of doping speculations about Chris Froome. I get it, we’re all skeptical of disparities in performance now, as if we expect 5 or 10 riders to all slug it out on every climb before sprint to the line mere seconds apart. When has that ever happened? Never! Yet, one rider attacks and leaves the rest of the peloton behind and he’s doping. Worse still, is all the arm chair analysis of riders using mysterious leaked data, or worse, climbing gradients, rider (guesstimate) weights, and climbing times. In my latest blog, I will review the PRO’s and CON’s “arm chairing”, why power is not the holy grail to stopping dopers, and most importantly, why you should largely ignore climbing calculations, like those made by Antoine Vayer.

Pro Cycling: Drugged up or Drug-free?

This year it was Chris Froome, but Nibali had numbers on par with Froome’s 2013 Tour. And why was no one questioning him this year? Too much bad luck, or just not spectacular enough? Lots of questions, but very few answers, and even less evidence. And no matter how much data Sky releases, it will never satisfy everyone.

On the other side of the isle is drug-free crowd. No one seems to doubt Tejay VanGarderen is clean, yet he was sitting close to Froome before he pulled out due to illness. Is this really fair? I mean I am no more confident that Tejay is clean as I am Froome. Froome is better, but he’s also more mature and at his peak. And Froome was far less dominant than everyone was trying to claim in the effort to pin him as a doper. Nonetheless, the UCI has learned nothing from it past. I know this because the great reformer, Brian Cookson went so far as to say:

“I think we saw riders during the last Tour de France were very tired and one of the causes, in my opinion, is the increased efficiency of doping controls…No one likes to see someone exhausted, but I think that this is a demonstration that we have constantly tightened the meshes of doping controls.”

As I commented on Twitter, fatigue doesn’t prove shit. Maybe riders use less, or rely on crappy drugs. Sure, the testing in cycling is much stronger than other sports, but making such statements is ill-informed at best. However, no other metric is used to prove or disprove doping than power output.

The Power of Power (output)

Suffice it to say, much has been written about the good, the bad, and the ugly of power. It’s not perfect and its easy to manipulate, or just mess up. Much has been written on the topic of precision and accuracy on this subject, so I will not be labor this point. The short answer is that power is a useful metric for monitoring performance only if the device is both accurate and precise. Assuming your power meter can be trusted (Stages, perhaps not), then it must be calibrated regularly. You must also assume that the power meter itself and the data have not been altered to provide false numbers. CycleSport ran a great piece on Thibaut Pinot, revealing a level of transparency unheard of in cycling.

Obviously, we could easily dismiss Pinot’s data as “junk” like the rest, but the power profile tracks very well with his development from Junior to Pro Tour rider, with no glaringly obvious leaps in performance. Such a profile is essential, in my opinion (and others), to creating a robust performance profile that combines the physiological and (bio)mechanical. Currently I do not see any of this, not from Froome, VanG, Nibali, or convicted doper Valverde – where was talk of his performance? It is too easy to cite “proprietary knowledge” (see Julien Pinot’s Opinion). We do not need training programs, we need the basic data over years. Is this type of data even available for all riders? Likely not until they turn pro. So if we cannot gain access to longitudinal data, can we compare past and present climbing numbers to catch the dopers? The answer there is complicated. 

Julien Pinot makes a case for longitudinal transparency.

Julien Pinot makes a case for longitudinal transparency.

Catching Dopers with Math?

The most widely used, abused, and myopic method for creating doubt are mathematical models using variables like gradient and speed to estimate climbing power output. Nothing makes for a great headlines, but the science behind those estimates is shaky at best. In fact, French researchers Grégoire P. Millet, Cyrille Tronche, and Frédéric Grappe did just that in 2014, publishing it in the International Journal of Sports Physiology and Performanceits worth noting that Prof. Grappe also coaches Thibault Pinot’s FDJ team.

Sixteen well-trained cyclists completed 15 randomized climbing trials n a range of slopes from 1.3 – 6.3 km and 4.4 – 10.7% gradient. They rode their own bikes equipped with SRM power meters both alone and in small to large groups and in seated, standing, or alternating positions. In other words, they not only accounted for the impact of environmental conditions, like wind, on climbing performance, but also how groups further influence climbing estimates.

Less Power than we think

I will skip over the statistics (summary table is below) of the paper and cut to the major findings:

  • Measured power is strongly correlated with Estimated power, with few differences when wind is low and gradient is > 4.4%. Estimations were not improved with steeper grades.

Estimations under non-windy conditions are + 6%

  • Factors related to aerodynamic drag, including inaccurate weather conditions like wind speed of rider drag are the most significant factors for increasing errors in estimation. This is exacerbated as wind increases, while slope does not appear to influence random errors.

Estimations under windy conditions are + 10%; that’s 40-45 W for Pro Tour riders.

  • Estimations of average power output for a group of cyclists climbing uphill are more accurate and “safer” to use.

Estimations for groups generally have errors of 1% or less even under windy conditions.

  • Estimating any individual power output value even under the best conditions are not sufficiently accurate to make any meaningful argument about a riders capabilities.

A realistic estimation error of 25 – 50 W should be noted in any estimation.

  • This study failed to account for rolling resistance, estimated body mass*, drive-train efficiency, estimation of hill gradient, and total distance traveled, all of which add more potential error.

* It is well-known that published body weights for riders are inaccurate, likely on purpose.

When we evaluate research we’re often looking major changes or the “AH HA!” factor. What we see from this study is mathematical estimations of power output for individual riders during competitions is at best a ballpark, and in my opinion, irresponsible at best. As the authors the authors note:

The current study underscores that it is impossible and dishonest to make comparisons between different cyclists and to release individual values such as PO without the corresponding range of random errors…as exemplified in the media- reported values for PO calculated from Tour de France ascents and presented as accurate (without any mention of the confidence interval). These values are produced to make comparisons between individual cyclists and feed the debate about the ongoing prevalence or decrease in doping in professional cycling. In our view, such comparisons between individuals are inaccurate.”

A more effective means for monitoring is the use of longitudinal power profiling. However, even here, we must assume that power meters are correctly and regularly calibrated. Moreover, we still do not have enough information to make any definitive assessments of what is “normal” or “physiologically possible”. Much has been made of past statements regarding the magic 7 W/kg as only attainable through doping. While this appears to be the case, it too would be inaccurate to assume that this is the upper limit of natural human limits.

Regarding specific comments about Chris Froome’s racing tactics, in particular his trademark attacks, specific training yields specific results. It is clear that SKY trains Froome for such specific attacks. To put it into perspective, legendary strength coach Al Vermeil once responded to a journalist’s question about how an NBA all-star on his team was able to make an astounding shot from one leg. Al’s response should be ingrained in any coaches memory, “Because he’s been there!” This is not to say that that specific move was practiced, rather, to highlight how important specific and complete training is. So when a journalist asks how is it possible that Chris Froome can spin his legs so fast, you need to realize, he has already been there many times before in training!

Written by