Froome's SRM data on Ventoux

[Moose McKnuckles]

Re: Re:

I think people are missing the really interesting thing here. It's not whether this data shows doping or not. It's that this data matches almost exactly the w/kg calculations of Vayer and Ferrari(?), even though Brailsford dismissed them as peseudo-scientific rubbish.

Exactly, there are two views on statistics. My background is in machine learning so I'll call them the Chomsky view and the Google view, as described here http://norvig.com/chomsky.html. The Chomsky/Brailsford view is that you have to perfectly model every conceivable factor with real physics etc and the model has to mirror reality. The Google/ammattipyoraily view is that the predictive accuracy of the model is what matters and not the internals.

My view is that the Chomsky way leads to whataboutery and nothing ever being done. If you have a model and it has very high accuracy despite leaving out certain variables, I think it's obvious that those variables aren't important.

I'm very much of that view as well. Often, what you're after in the lab is a parametric response. As I turn this knob, what happens to the output - that kind of thing. Use that data to create a model. Working from first principles to write a deterministic equation from which to model your data is a great big time suck and that's about it.

John Swanson

I'm shaking my head at the sheer amount of BS here. You guys are both spouting nonsense.

First of all, just because a model has a lot of accuracy doesn't mean it is a) replicable outside the sample, or b) it is a good model. A model may be used for a) prediction or b) explanation. A model that does b) doesn't necessarily do a). A key example of that is a model that contains a high degree of multicollinearity.

That "two views on statistics" comment? Complete and utter crap. Look, you are probably 100 times better than me at machine learning, but the stuff you are saying about statistics is nonsense. 100% crap. There is no "Chomsky" or "Google" view. Good grief. Go read about Fisher, Gosset, Pearson, or Neyman.

Second, Swanson, you shouldn't be agreeing, because he says exactly the opposite of what you are saying. He's not talking about "deterministic" models. Those by definition have 100% accuracy, because they have no randomness. That last sentence you wrote is sheer nonsense just for the sake of sounding intelligent.

Don't pull that stuff. You're good at what you do, so do it. And stick to it.

 

[Escarabajo]

Re:

The easiest way to fabricate the data would be if you had the real data for that climb from someone else and then apply a Froome Fudge Factor. Vayer posting it as the real data would indeed leave him open to ridicule if it's not, but I don't see how he could have possibly gotten his hands on Froome's data. Surely Sky Science is hidden in some super max security vault.

+1

It is simple. To fake it is easy. Anybody can do that. To fake it with real numbers is the problem. You have to be perfect to do it.

 

[Tonton]

Re: Re:

The easiest way to fabricate the data would be if you had the real data for that climb from someone else and then apply a Froome Fudge Factor. Vayer posting it as the real data would indeed leave him open to ridicule if it's not, but I don't see how he could have possibly gotten his hands on Froome's data. Surely Sky Science is hidden in some super max security vault.

+1

It is simple. To fake it is easy. Anybody can do that. To fake it with real numbers is the problem. You have to be perfect to do it.

Vayer wouldn't compromise his earnings for a stunt. He already got his niche as the TdF master of cleanliness.

 

[Escarabajo]

Re: Re:

Solution to all this:
Power meters and HRMs(that riders can see) should be banned from racing anyways.
They add nothing for the fans, and take away many facets of racing.
If not banned they can use a PM on the condition that the figures are made public.
Would only be fair.

Power meters don't help you go faster. It only measures what you have done or are doing.

Doping helps you go faster and should be banned.

To all the people out there that are fighting the good fight to keep doping legal
What are you people even on about?

HRM and powermeter don't help anyone go faster derp. No help for staying under threshold herpderp

You are being too defensive and you don't post much in this forum either. Not a good combination of objectivity.

 

[HiroAntagonist]

I've got to agree on the HR being a somewhat inadequate indicator. It's more like a gloss over akin to say BMI, and really isn't built for an accurate overview of widely varying individuals. It may point at something suspect, but on its own...

 

[Microtom]

Low heart beat is very normal for an athlete such as Froome, even in a situation of extreme physical activity. It is caused by exercise-induced cardiomegaly.

 

[More Strides than Rides]

Re: Re:

Sorry HR info is a furphy.

Furphy: is that how you say "funny" in Australia? Furry? Fluffy?

I had to google it. It means "rumor".

Vayer and Ross Tucker are vouching for it's authenticity, each saying to have have the raw data. One of them made the point that 2800+ data points would be hard to fake, and they do line up so well visually with the video.

I really want to see where this story goes. The leak is ine interesting story: the who's, why's and how's. The numbers, assuming accuracy, speak for themselves. The science though, doesn't.

It's not that HR is not understood well, it's just that it's relationship to other physiological pieces is not. Opd Hr training zones had little connection to actual metabolic states, and even those are not discretely aerobic/anerobic like people want to believe. Same thing with regards to VO2 max or % of VO2 max. Or muscle tension. Or muscle fiber recruitment. Or the psychology of perceived effort. Or heat management... All of these factors that themsleves are one or two steps removed from performance.

So the discussion here about %max heart rate just isn't the right one to have. The science is really far behind on really knowing how it all cones together.

Back to the interesting storylines though, is his static heartrate during his attack. Whatever the number is, it should change. My intuition says that even thinking about attacking should change it. Unless there is a HR monitor machine calibration error, the only explanation is a motor.

 

[Dear Wiggo]

By furphy I mean it's not the flag in the video / file that indicates the problem. It doesn't indicate - to me - doping. Max HR, etc, is affected by a number of things, including fatigue. His HR response based on his HR from the 2011 TT looks pretty normal to me.

 

[DirtyWorks]

Re:

Low heart beat is very normal for an athlete such as Froome, even in a situation of extreme physical activity. It is caused by exercise-induced cardiomegaly.

Can you clarify? "Low heart beat" equivalent to "low resting hearbeat?"

I am from the Polar heartrate monitor era.... Guys had all kinds of max heart rates. Some low, some high and no correllation to power.

I too am fascinated by the "leak." A bizarre decision.

 

[durianrider]

The video link doesnt work. Please update it.

Where is the video of Froome and Vayer?

HR means jack all. Froome doesnt even take caffeine when racing according to his book so it would be natural for someone so fit to have a lower HR if there are no stims present. With typical dosages of stims he could spike his HR at least 15-20.

Cheers.

 

[Blakeslee]

Hopefully there is more information coming, as is indicated in this article (http://www.cyclingnews.com/news/vayer-says-froomes-supposed-ventoux-data-was-leaked-not-hacked). It was troubling and discouraging watching Sky's performance in today's stage. Hopefully there will be credible data released that will start the dominoes falling.

 

[Dear Wiggo]

Froome doesnt even take caffeine when racing according to his book

Sure, and everything he tells us is the truth, the whole truth and nothing but the truth, right?

Did he mention the asthma he has had since childhood in that book? No? Huh. Weird.

 

[enCYCLOpedia]

The video link doesnt work. Please update it.

Where is the video of Froome and Vayer?

There is no video of Froome and Vayer. There is a video that overlays the 2013 Ventoux stage with Froome's power and HR data. It has been taken off youtube, but L'Equipe has embedded a link to a new version in this article:

http://www.lequipe.fr/Cyclisme-sur-route/Actualites/Chris-froome-sky-n-a-pas-seme-la-mefiance/574244

Absolute HR values are meaningless. However, they are meaningful in context. And, of course, the power data are meaningful, as they confirm the VAM calculations - Froome is not miraculously efficient; the power data for Froome lines up well with the calculations of the "clowns".

Of course, to think that Sky was "hacked" is ridiculous - did they keep it on the teamsky.com web server? The data has clearly been leaked, which means someone at Sky wanted it to get out. And which is why Dave Brailsford doesn't appear to want to talk about the "hack" anymore.

 

[Mr.38%]

Is Gesinks power avg. on Strava actual power meter data or calculated based on his alleged weight and calculated/statistical data (Strava's algorithm ain't bad). We really need more riders to publish their actual data, so the pseudo scientists could do their magic.

Froome's low HR appears to be inline with former data available from TrainingPeaks however.

 

[Cycle Chic]

Re: Re:

Back to the interesting storylines though, is his static heartrate during his attack. Whatever the number is, it should change. My intuition says that even thinking about attacking should change it. Unless there is a HR monitor machine calibration error, the only explanation is a motor.

This !! ....and thats why Froome said he would never have a yellow jersey taken off him.

 

[danielovichdk2]

Am I the only missing more files or data from the so called hack ?

I absolutely don't believe no one was hacked until I see the files from a source that has some credibility.

At this point it might as well have been a leak from l'equipe and vayer trying to scare sky, no ?

 

[Alex Simmons/RST]

Re: Re:

Nice analysis @veloclinic in terms of W' balance: https://twitter.com/veloclinic/status/620827797284618240

Hard to believe Froome is working all that out in his head as he rides, but does explain the constant PM watching. Do the SRMs have W' as a field?

No they don't, certainly not the Powercontrol 7 that he would have used in 2013. My understanding of the PC8 is such a feature is not available in it either. Such power-duration modelling hasn't quite made it to head units yet. In the years ahead perhaps (although not sure it'll be all that useful from a race POV).

Sky don't currently use SRMs in races in any case. They've been sponsored by Stages for last couple of seasons.

This isn't the place to go into W' or other power-duration model stuff, especially given people have trouble with something as basic as heart rate.

 

[Alex Simmons/RST]

Re:

Keep in mind fatigue also lowers your HR. These are stage 15 or 9 where we get to see Froome's HR. It's going to be down ~5bpm at max effort from fatigue alone. Add in a lower MHR and this kind of thing is not unusual.

Max HR also falls with increasing fitness / as stroke volume increases.

 

[Alex Simmons/RST]

Re: Re:

4. Suggesting this data is confirmation of the accuracy of a power estimate made from climbing speed is ignoring the quite possible bias error in both estimates due to unaccounted for errors. Two data wrongs don't make a right. I'd expect better from scientists.

Bias error from what? How much? This isn't a single data point: estimates have been compared with actual data from cyclists who voluntarily released their files before and have been within a % accuracy. Whereas you're saying the whole thing is junk based on what? Unknown factors of unknown magnitude? I hope I never get a reviewer like you on a paper.

I went through that potential bias errors earlier in this thread. If you want to be thorough as in science, you should always question your data, ask the question - "could this data/model/estimate be wrong?" Then set about finding ways that might show it is wrong/not valid/doesn't work/or has limits on validity.

Only when you've exhausted avenues of showing something is wrong can you then begin to have greater confidence it is right. I hope you do get such a reviewer, because then you'll learn to be thorough and have a robust approach to research.

If a scientist is knowingly using illegally or unethically obtained data, they have zero credibility IMO. I'd expect any credible scientist to distance themselves from those that obtain their data in this manner.

But you don't know how the data was obtained. So your view is by merely examining and commenting on the data you lose your credibility. Convenient.

Perhaps you missed the bit where I put an "If" at the start of my sentence.

 

[Alex Simmons/RST]

Doctors, I'm confused. So if Froome had raced the 2014 Tour like Ventoux then no podium for him??

Power data needs duration and other contextual information to make sense and determine which, if any, make for reasonable comparison.

 

[Alex Simmons/RST]

Re:

I'm trying to work out a way to plot the data from those tables in a sensible way that incorporates the length and steepness of the climbs. I made a couple of very simple plots of the averages and posted them here before which was interesting but was definitely skewed by very long and very short climbs (and also road surface in one case). I intend to do it when I have less work and more time.

My personal view based on nothing other than intuition is that the Ferrari formula underestimates watts/kg but the important thing is that we're always comparing like with like so it doesn't matter so much.

First of all assess which are reliable indicators of a rider's mean maximal power (versus a rider not being at their limit).

Then plot power v duration. Then colour code context information.

 

[42x16ss]

Re:

Am I the only missing more files or data from the so called hack ?

I absolutely don't believe no one was hacked until I see the files from a source that has some credibility.

At this point it might as well have been a leak from l'equipe and vayer trying to scare sky, no ?

Yeah probably, just like Armstrong's positive tests...

 

[Alex Simmons/RST]

Re: Re:

Solution to all this:
Power meters and HRMs(that riders can see) should be banned from racing anyways.
They add nothing for the fans, and take away many facets of racing.
If not banned they can use a PM on the condition that the figures are made public.
Would only be fair.

Power meters don't help you go faster. It only measures what you have done or are doing.

Doping helps you go faster and should be banned.

Doping is already banned.

 

[armchairclimber]

So, to summarise, the data that was overlaid was Froome's data obtained by Vayer. That data shows an average of 5.88 wkg for the climb, which is not particularly suspicious for the stage winner. His HR shown on the screen peaks in the low 160s. This is contextualised by Froome's assertion that he can't get his HR above 170. Given that he would be fatigued and that sharper peaks in HR will have been "smoothed out", there doesn't appear to be anything untoward.
It looks to me as though Vayer has done a pretty good PR job for Froome. I wonder if the data was given to him deliberately.

 

[Pies&Booze]

Anyone read The Times today? Apparently Froome says the numbers from Ventoux in 2013 can't be used to prove doping as "That's nuts, especially seeing as the data in question is over two years old".

OK that makes sense.........

At least The Times is talking about the issue.