The Science of Sport blog has been one of the leaders in charting climbing speeds and power over the history of the Tour. This year, with more riders publishing their daily power files during the Tour, the data is getting more and more precise. I spoke with Ross Tucker about some of his findings.
Andy Shen: How’d you get interested in estimating wattage from climbing speeds? Was it via Ferrari or Vayer?
Ross Tucker: I got into the power stuff because:
a: We do it all the time in our research lab – we test a lot of guys, at a range of levels, and being a physiologist, the limit to performance is interesting anyway.
b: I met David Walsh in 2009, when he was out in South Africa. Had dinner with him and spent a day hiking on the mountain. A fascinating guy, and he mentioned that he’d tried before to quantify performance and determine whether there was a physiological limit that might be able to expose doping. He introduced me to the numbers of Portoleau and Vayer (gave me a little french book they’d written – I understand nothing except their tables of times and powers). Then I picked it up from there. And in 2009, when that story came out about Contador having a VO2max of 99.5ml/kg/min, it was an obvious thing to pursue!
AS: Speaking of Contador, some say the calculations on Verbier were wrong, others say the effort is shorter, so he should be able to sustain a higher wattage. Do you think Contador’s performance on Verbier is definitely indicative of doping?
RT: No, not definitely, because of the reasons you mentioned about the assumptions one must make over the length of the climb and the conditions on the day. The method of assessing performance using overall performance time (or VAM, which is really just the same thing) is imperfect for the simple reason that the performance is affected too much by conditions. Power output is not, and that’s why directly measuring the power output, as has happened in the last two years, is such an improvement.
Back to Verbier, the climb lasted 20 minutes, and there were reports of a following wind. Both factors would drive the estimated power output up – the estimate will exceed the actual power output because of the wind, and the shorter length will allow a rider to produce a higher relative intensity than on a climb like the Tourmalet this year, which took just under 50 minutes. So no, it’s not proof of doping.
AS: Do you think we saw a clean and less Ricco-like Contador this year, or do you he was less explosive this year due to bad preparation or illness?
RT: I think he was slightly below the level he was at last year, but not by the enormous amounts that people seem to be suggesting. I would say that his variation in performance is within a normal variation in performance from one year to the next, especially over such a long event. You see this all the time, when guys have great form one year, but not the next – Wiggins is an example of a particularly large drop in form. The same happens in other sports – marathon runners will follow the same preparation and produce times up to 3 minutes apart. They’ll threaten world records and then six month later, fail to finish at a significantly slower pace. Preparation is not an exact science, and so of course, Contador may well have been a shade below his level from last year. Then of course Andy Schleck was better, and the result was a competitive race where 2009 was a controlled victory.
What is missing in all of this, of course, is precise measurement of performance and power output. Contador’s estimated power outputs are not dissimilar from previous years, incidentally. The Verbier apart, he has rarely climbed at much above 6W/kg (this is the estimation, mind you), with the exception of the Montee Jalabert, but that was only 10 minutes long, and like the Verbier, not directly comparable.
So I would love to answer your question and say "yes, he was cleaner", but the reality is that his other preparation could just as easily have been the difference between 2009 and 2010. I think for now, I’d be satisfied to say that on the whole, the Tour seems to me to be slowing down, and there are signs of a return to performance levels that we did not see in the 90’s and 2000’s. I hope that this is true from the yellow jersey all the way to the Lantern Rouge, but only time will tell. And of course, the accurate measurement of power output rather than estimation, which is certainly valuable and informative, but not fool-proof or exact!
AS: A common argument I hear often against using performance ceilings as indication of doping is that you can’t account for outliers and exceptional performances. Usain Bolt is often used as an example in that argument. Is cycling different from running in that we probably won’t see a clean athlete do 6.2+ W/kg?
RT: I believe this, yes. I wouldn’t want to make it appear as though there is a distinct ‘line in the sand’ where a performance instantly becomes "impossible", because physiology is not binary – it’s not "on vs off", "black vs white" and so it’s certainly not as simple as saying that a performance of 6.1W/kg is a clean performance but a performance of 6.2W/kg is a doping performance. In a discussion around these matters, you have to kind of commit to a value, even if you talk about ranges, but they’re not definitive lines where you cross over from realistic to impossible in an instant.
What I think is true is the following: There is a physiological implication to every performance, and if you know the performance then you can begin to appreciate whether the implications of that performance are realistic. And given the history of the Tour, the estimates of power output from the likes of Vayer, Portoleau and even Ferrari (in his own words), cyclists were regularly able to sustain power outputs of 6.3W/kg, 6.4W/kg, even 6.6W/kg on the finishing climbs of the Tour lasting 40 minutes or more.
Now, in 2010 (and 2009), you see the same assumptions and estimations, as well as SRM data, revealing that the very best riders in the Tour are riding along at 6 W/kg. And yes, tactics play a role, and yes, conditions on the day influence the ‘error’ in the estimates and even the physiological implications of your direct SRM measurements. But over the course of perhaps a decade of Tours in the 90’s, and a decade of Tours in the 2000’s, and up to 30 climbs per decade, this error starts to be reduced ‘in the mix’ so to speak.
And this leaves me, anyway, with the thought, or hypothesis, that performance in the 2009 and 2010 Tours have been curtailed. The 6.3W/kg finishing climbs of the 90’s and 2000’s would have seen Contador and Schleck finishing minutes down on a climb like the Tourmalet. In other words, those EPO/blood doping era riders would be to Contador what Contador currently is to Gesink, Menchov, Sanchez. Is the standard of cycling that much lower? I don’t think so, and my thought is that so far, we’re seeing an overall suppression of doping in the Tour.
Not its removal – you and I both know that when so much is at stake, and there is such a pervasive history of doping in the sport, you don’t just throw a switch and hey presto, it’s clean. I’ve no doubt that there is doping, probably very sophisticated, perhaps driven to fly below the radar now imposed by the biological passport system. Whatever you make of Floyd Landis’ allegations, for example, you have to take seriously his explanations of "micro-doping", of how riders were able to avoid detection through reduced levels of the same practice. When Prof Ashenden, whom I have a lot of respect for as an authority on doping and cycling, says that this explains what they are seeing, then I think you have the evidence that should shatter any illusions of a "clean Tour".
I believe it is cleaner, that the biological passport is having an effect, but as in life, there are people who will attempt to beat the system.
Just to comment on the outliers like Bolt – it’s interesting that the statistical modeling of athletic performance actually suggests that the 100m world record still has about 0.1 seconds of ‘reserve’ in it. These kinds of mathematical and statistical models always gain momentum around the time of the Olympic Games, and so leading up to Beijing in 2008, a few different groups modeled World Record times and world leading times per year. This is obviously not possible for cycling as a result of changing conditions and the fact that times are of less importance than victories.
One group from France found that most records had stagnated, or would within the next 30 years, while another suggested that the 100m WR should drop to somewhere in the 9.4s range. Usain Bolt moved the 100m event forward by a few generations in one year, and so of course, there is suspicion about him (as there is guaranteed to be given that the previous Olympic Champion and just about every WR holder before him was implicated in doping). But, the point I would make is that his performances defy only our historical perception of what is achievable. It is very difficult to measure the physiology of a 10 second sprint, of course, and so the same approach as I’ve tried to use for cycling is much more difficult to do for 100m sprinting. But my argument here is that Bolt is an astonishing athlete, generations ahead of his own time, but he’s not doing the impossible, physiologically and statistically speaking.
AS: Another argument (and this is the really stupid one), is that people do extraordinary things in extraordinary situations – the “parent lifts a car to save a trapped child” story. Endurance sports don’t work like that, do they? Can you do something in a race that you couldn’t do in training?
RT: As you say, this is not how endurance sport works. Yes, it is true that the body has a reserve, and that this reserve can be accessed in very specific situations. We know that the brain regulates exercise intensity by controlling how much muscle it allows one to activate, and that this regulation exists specifically to protect against potentially harmful physiological changes. For example, it’s now pretty well established that when you cycle or run in hot conditions, your body has a "fail-safe", a short-circuit switch at around 40 to 41 degrees celsius. As soon as you reach this level, exercise becomes impossible to sustain. The symptoms are dramatic and obvious – readers might look up the story of Gabrielle Andersen-Scheiss in the Los Angeles Olympic Marathon in 1984. Or Jim Peters in Vancouver in 1954. These are dramatic illustrations of "failure" of the body to regulate intensity, and they’ve hit this body temperature, which has been called the critical level of hyperthermia.
Now, physiologically, the body (through the brain) is too clever to allow this to happen, and the reason that athletes slow down in the heat is precisely so that they don’t overheat before the finish line – staggering to the line like Peters and Scheiss is not a great strategy for optimal performance! It happens because the racing situation forces athletes to approach the critical level. But the point is that it doesn’t produce an optimal performance – exercise is regulated in advance of this kind of physiological “meltdown”. The same applies for energy supply, for oxygen delivery to the muscles, to muscle pH levels, plasma osmolality. All are regulated in advance of some potentially catastrophic failure (and of course, sometimes it does fail, and people faint, black out, drink too much etc. But these are pathological conditions, with extenuating circumstances. Think of how often you’ve seen an elite athlete in serious trouble after a race. It hardly ever happens.)
But the "reserve" capacity is a consequence of this regulation – there is always a reserve, a degree of muscle activation that we don’t use because we can’t. So when one argues that the body is able to over-ride this reserve, then they are rewriting physiology. It’s like saying that someone can commit suicide by holding their breath – physiology wins 100% of the time. Picking up cars? For efforts lasting seconds, yes, this may be possible within the realms of physiology. But over hours a day, for three weeks? Even over 40 minutes? There are too many built-in safety switches that get you first, so no, you don’t have two parallel physiologies. Of course, the race situation takes you closer to the limit than in training, and so yes, you do excel relative to training or laboratory testing. But it’s the same physiology and I do believe that there is a ceiling.
AS: Now that more and more riders are posting their power data, have you been able to check your calculations to see how accurate they are?
RT: Not directly, back to 2009. What has been interesting is to take someone like Chris Horner, for who we have SRM data, and then to make the same estimates for him as have been made in the past. For example, his SRM reveals that he rode the first half of the Tourmalet at say 5.9W/kg. If you make the VAM or power output estimate calculations for this segment, you find values around 5.8 W/kg, which is a pretty comforting comparison. On other occasions, the SRM data have provided power outputs that are slightly lower than the estimates using VAM or other equations. This is to be expected – sometimes a rider is drafting and thus going faster with a lower power output, other times they are into a headwind, not drafting, and thus going slower with higher SRM power output.
But all in all, I’ve not seen anything in this Tour de France of 2010 that makes me dismiss VAM or estimations. They’ve always been within 0.2 W/kg, usually less, of the SRM values, and that means that the estimation method of Vayer and Portoleau is probably more accurate than many people are giving it credit for. Similarly, the VAM method, provided the timing and altitude changes are accurate, produces values that are at the very least intriguing, and at the very best, compelling to support the discussion.
AS: Do you foresee a future where ProTour riders are tested for their VO2max, and have sealed SRM’s on their bikes?
RT: Tested for VO2max, no, probably not, because it’s too unreliable in that the rider’s motivation, their state of training, the phase of the season etc are going to affect the result. Even the mass co-ordination of this kind of testing is problematic because of differences in equipment and calibration from one place to the next. And I can’t see one laboratory being designated as the testing center for perhaps 300 pro cyclists a year! Too large scale and too many errors.
I would love to see more measurement of power output. Here there are problems too – calibration, for one thing. Then, I can appreciate that the top cyclists don’t wish for this info to become publicly available. Alberto Contador, for example, might squirm at the thought that his power output data from 2010 can be compared to his values from 2009 to reveal potential weaknesses – maybe his attacks lacked power output by 10% compared to previous years, and suddenly, he’s shown, through the numbers, to be vulnerable.
But, I don’t see this as insurmountable. In fact, I’m all for openness and transparency. Perhaps the data is held for 3 months after a race before being released. Perhaps it is held for a week or the duration of the race.
The other thing is that people tend to overstate the importance of having this kind of information during a race. If Andy Schleck knows, for example, that Contador is producing 10% less power in those 20 second attacks and may therefore be vulnerable, so what? It’s not as though Schleck can suddenly tap into a hidden 10% that he has in order to beat Contador the very next day! Knowing someone is worse than in 2009 is completely meaningless on the road, when you are riding side-by-side and cannot find the power to crack them! And if Menchov is being dropped by those attacks, what does it matter whether they’re at 700W or 780W? Either way, he’ll know he can’t respond simply because he can’t respond!
There is this ingrained belief that science will give a rider an instantaneous edge on the road, and I don’t believe this to be true – it’s far more nuanced than that! The scientific data provide a good insight in hindsight. And as a scientist (and general number junkie), I think they’d add a lot of value to the coverage of the sport, and that would help it grow! Athletics finds some of its appeal in the measurement of performance, and while cycling is a different beast, measuring it doesn’t mean you detract from the tactics, the battles and the "romanticism" of the sport.
Back to the analysis, when you think of the invasive nature of blood and urine testing, that cyclists are now going to have their blood stored for years for retrospective analysis, that experts are scrutinizing changes, it seems to me that the measurement of power, if kept confidential, is less invasive, and certainly less likely to blow a doping case open. I do think that it provides some context and insight to performances. And the value is mostly in tracking a rider from one race to the next, from one year to the next.
I dare say someone like Ricco would have jumped out as a serious red flag, for example. His performances will have been noticed simply because he was winning and yes, that would invite the testers to scrutinize him anyway. But if I’m one of those testers, one of the scientists on the biological passport team, I can look at the sport, with objective measurement and get an idea of how it is evolving, how our interventions are impacting the sport and if I see a rider who suddenly appears with performances that one might deem "grey", then you ask whether that rider is possibly doping. The biological passport data takes on a different meaning. It’s not conclusive and it will never constitute proof, but it provides context, re-assurance (to some degree) and insight. So I’d love to see it.
AS: Some of the most interesting data from this Tour was Anker Sorenson’s numbers, where you got a unique insight into the strategies on the climbs. I found it really interesting that Anker Sorenson’s pull was actually faster than the subsequent battle between Schleck and Contador – the softening-up was harder than the actual head to head fighting. (It sure seemed the reverse was true during the Armstrong era.) I think this is definitely an example of data adding to the our experience as fans, not, as you say, taking from the romanticism.
RT: This was interesting, yes. I will say that if you compare Horner’s data to Sorensen’s, the difference is not that large – this is the benefit of being in the pack, the drafting effect, because if you look at the data, Sorensen is producing about 6.5 to 6.6W/kg for the first 10 minutes of the climbs, while Horner, riding at the same speed is around 20W, or 0.3W/kg lower. So he’s at 6.2 to 6.3W/kg for a short time.
But the key, as you point out, is that this was not sustainable. Not for Sorensen (which will not surprise people – as good a rider as he is, he’s not playing the role of a GC contender), not for Horner, and, it would seem, not for the elite top 5 of the Tour either. The Tourmalet illustrated this very clearly. The group was broken up on the early slopes, and after the attack, about 10km out, the gap was created and then stayed at around 90 seconds. So the effort to create the gap, even for Contador and Schleck, was not sustainable. This illustrates that those efforts came at a cost, a cost that could not be paid over the entire 40 minutes, even when the race demanded it (on the Tourmalet).
Quite how this compares to the Armstrong era, I’m not sure. It would certainly make for interesting analysis. It did seem that he was faster, but then if we didn’t have the comparison between Sorensen and Horner and co, we might have suspected the same of Contador and Schleck. The time gaps are very interesting though. The days of attacks near the bottom of the climbs, 10km out, with a lead growing steadily now seem to be gone. Even last year on the Verbier, Contador created a gap of 30 seconds within 3km, but then only 15 seconds over the next 3, and then Schleck pulled time back, over the admittedly less steep sections at the top. The fact that in this year’s Tour, Contador and Schleck finished the Tourmalet’s final 2 or so kilometers riding at pretty much the same power output as Horner, around 5.8W/kg, is telling, however.
AS: One great thing about all this is that cycling fans are becoming more and more sophisticated. I daresay the days of people like Chris Carmichael saying that 6.8-7 W/kg is what it takes to win the Tour are over.
RT: Yes, I think so. The implication of 7 W/kg for any length of time is significant. Given that we have some indication of Lance Armstrong’s cycling efficiency from the research done on him by Ed Coyle, it’s possible to estimate the oxygen cost of riding at 7W/kg. Coyle found that back in 1999, Armstrong’s efficiency was 23.1%. Using this measurement, 7W/kg would "cost" about 86 ml/kg/min of oxygen. Now, if a cyclist can sustain that for even 30 minutes, then they’ve got a massive engine, the likes of which have never been seen, not even close. An estimate for VO2max here is somewhere between 96 and 101 ml/kg/min. The analogy is that if a state-of-the-art supercar suddenly found itself riding alongside a Toyota Prius, you’d wonder what was under the hood of that Prius! Even if you raise the efficiency, you’re left with highly unusual physiology.
And I don’t want to fall into the trap of having a "reductionist" approach to cycling performance – the VO2max is one of many variables to affect success and performance. It would be an error to say that the guy with the highest maximal capacity wins – we know this is not true, because efficiency is very important, and conditioning allows a cyclist to approach that capacity for longer at the end of long rides. But the combination of physiological attributes that would be required to sustain say 500W or 7W/kg over 12 or more kilometers at the end of a 5-hour day in the Tour? In my opinion, questionable. And the 2010 data provided by the SRM suggests (doesn’t prove) that this may well be the case – we saw sustained power outputs of 6 W/kg on the longer climbs, and that’s 15% lower than this so called ‘magic number’ thrown around by coaches.
Some have argued that the assumptions I’m making are far too conservative, that cyclists could easily have the physiological attributes to produce these numbers. Well, 2010 certainly didn’t have any. Nobody even came close to 6.5W/kg or more, except on a 10 minute sprint up to the summit of the Montee Jalabert. Therefore, if you did the exercise in reverse, and used performance to derive physiology, you’d arrive at the same "assumption values".
Interestingly enough, Michele Ferrari has often spoken of a 6.7 W/kg requirement to win the Tour. This was documented in Armstrong’s own book and by others (Coyle), and I have no reason to think they’re making it up. Ferrari was clear in interviews where he said that 6.3 W/kg is not enough to win the Tour (this was during Armstrong’s time with him), and that "Lance needs to be at 6.6 to 6.8 W/kg". Those are extra-ordinary numbers, with extra-ordinary implications. As I said before, they would produce climbing times that are minutes faster than anything from Contador and Schleck this year. So one can guess at why the top riders of 2010 are apparently so much "weaker" than Ferrari’s magic number. Not proof of doping, sure, but it asks more serious questions of cycling’s champions in the 90’s and 2000’s.
AS: I’ve heard that Armstrong told a sponsor he did the Alpe d’Huez TT at 500 watts…
RT: I attended the meeting of the American College of Sports Medicine (ACSM) in 2005, in Nashville, about 8 months after the Alpe d’Huez TT. A student from Texas presented an analysis of that climb, with every single one of the 21 hairpins accounted. Of course, he was modeling power, not measuring it directly, but the approach was to factor in the changing gradients up the climb, and the wind speed and direction. This is arguably a more accurate method than simply taking total time, total distance and average gradient to calculate the power output. And his model estimated a power output of 495 W. Unfortunately, the presentation (which was a poster) never made it to a journal, so I can’t provide a reference, other than the conference proceedings book for 2005. Of course, this was a 40 minute effort without the 5 hour stage preceding it, but nevertheless, an extra-ordinarily high power output. Ferrari has suggested 6.7W/kg before, as I think I mentioned to you in the first set of questions. This is also equal to 500W. Unfortunately, we’ll never know exactly, but it’s an intriguing set of numbers in an interesting debate!