The Chris Squared Thread

[thehog]

Is there a video of this stage at Vuelta?

I'd like to refresh my memory and make my own mind up.

I've answered my own question.

http://www.youtube.com/watch?v=QoeuRAn7_6E&sns=em

2011 Vuelta full gas FroomeDawg

 

[thehog]

I've answered my own question.

http://www.youtube.com/watch?v=QoeuRAn7_6E&sns=em

2011 Vuelta full gas FroomeDawg

I've watched the video. Hard to say. Froome did a lot of work. Also Wiggins had a mechanical 5km from the Angulia and he and Froome had to chase back just before the climb.

Froome was stronger but was out front like Cobo for the entire climb.

Have to say its hard to use this climb as full gas Froome.

Sky got their gearing wrong as well.

Cobo 34×32 (28.3-inch);
Nibali 34×29 (31.2-inch);
Froome and Wiggins 38×32 (31.6-inch);
Kessiakoff 34×28 (32.3-inch);
Fuglsang and Mollema 36×28 (34.2-inch);

I'm not convinced to be honest.

It's not a good example of Froome and this was his breakout race. Froome 2013 is twice the rider as he was in 2011 - which is a scary thought.

Let me so some more reading.

 

[ulrichw]

The best SRM file, and weight, to have would be Quintana on the Semnoz. [...]

I'm curious about this portion of your statement:

One thing I don't think I've seen mentioned here (at least not in the threads I follow) is the ease with which any power files can be manipulated.

There's no technology in place to guard against forged files (at least not that I heard about). It would be trivial to programatically manipulate the files to modify any criteria you wanted to while keeping the data internally consistent. (e.g., making sure relationship of speed to wattage stays consistent)

If you wanted an attack to look not quite as violent, it would be easy to smooth that out. Longer periods might be harder to manipulate, since other references could be used to cross-check the data (e.g., measuring interval timings from TV), but there's still likely to be enough imprecision in these methods that there's enough wiggle room for substantial undetectable manipulation of the file.

 

[Race Radio]

I'm curious about this portion of your statement:

One thing I don't think I've seen mentioned here (at least not in the threads I follow) is the ease with which any power files can be manipulated.

There's no technology in place to guard against forged files (at least not that I heard about). It would be trivial to programatically manipulate the files to modify any criteria you wanted to while keeping the data internally consistent. (e.g., making sure relationship of speed to wattage stays consistent)

If you wanted an attack to look not quite as violent, it would be easy to smooth that out. Longer periods might be harder to manipulate, since other references could be used to cross-check the data (e.g., measuring interval timings from TV), but there's still likely to be enough imprecision in these methods that there's enough wiggle room for substantial undetectable manipulation of the file.

I have seen many people claim that SRM files can be manipulated but I have not seen a clear explanation how yet. A key element is to have multiple files from multiple riders, then it just becomes math

 

[DirtyWorks]

I'm curious about this portion of your statement:

One thing I don't think I've seen mentioned here (at least not in the threads I follow) is the ease with which any power files can be manipulated.

There's no technology in place to guard against forged files (at least not that I heard about). It would be trivial to programatically manipulate the files to modify any criteria you wanted to while keeping the data internally consistent. (e.g., making sure relationship of speed to wattage stays consistent)

What is trivial is smoothing and averaging up/down. For example, somewhere there's still a site rewriting gps track files for Strava glory. Not hard to do since the logic is practically global and therefore straightforward.

Hey look! It's the binary file format!
http://www.stephanmantler.com/files/SRM/SRM_File_Format_Specification.pdf

As someone that spends some time making things like "seasonal adjustments" to data, I can tell you what you imagine as trivial is not. Picking a point in time and smoothing out just that little chunk is a ton of work.

Race Radio,
They would do it by manipulating the data chunks of the binary srm file. PITA does not sufficiently describe the effort required to selectively modify data.
http://www.stephanmantler.com/files/SRM/SRM_File_Format_Specification.pdf

 

[Benotti69]

There is some manipulation going with riders weights. That riders are not honest about their weight is not on. If the riders weight is unknown does that not make the maths a guessing game?

 

[martinvickers]

There is some manipulation going with riders weights. That riders are not honest about their weight is not on. If the riders weight is unknown does that not make the maths a guessing game?

From the mouths of babes and sucklings, you have perfected praise.

OF COURSE the maths is a guessing game. That's the whole problem!

 

[ulrichw]

As someone that spends some time making things like "seasonal adjustments" to data, I can tell you what you imagine as trivial is not. [...]

[...] It would be an epic amount of work to selectively modify the gps tracks though and it would never be perfect anyway.

http://www.stephanmantler.com/files/SRM/SRM_File_Format_Specification.pdf

ok, perhaps I was getting excited when I used the term "trivial" - Let's just say that for someone with programming expertise (I'm a software engineer by trade) it's not too difficult to come up with a program/scrip that does this.

Admittedly, redoing the GPS tracks is a level more effort, but I would argue it's not that difficult either.

For example, what I'd do to smooth out an attack is to use the existing points starting and ending say 10-20s around the attack to create a path, fudge the power/speed to yield the same average speed over the entire segment (so the start and the end of the segment end up in the same locations as the original), and then interpolating new gps coords over the computed path. For good measure you could add a small random perturbation to the points to simulate GPS jitter.

I'll concede: this is not trivial, but this is not that hard, either - certainly not "epic."

 

[ulrichw]

I have seen many people claim that SRM files can be manipulated but I have not seen a clear explanation how yet. A key element is to have multiple files from multiple riders, then it just becomes math

Yes - cross-checking against a known reference would be one way to determine certain types of manipulation (for example reducing the average wattage over an entire ride). As a matter of fact, this is a possible alternate explanation for data you mentioned before -where Chris' weight computes to 61kg vs. 65. Another possibility is that his SRM data was averaged down a little to make it seem more reasonable, and his weight actually is 65 (not saying this was actually done - I do think it's more likely that his weight was overstated).

You could still get away with smoothing away bumps - for example the peak power reached during an attack. The only way to detect this type of manipulation would be to have the power data for someone who put out exactly the same effort at the same time (not usually the case for an attack).

There are also inaccuracies in power data - nominal accuracy for most power meters seems to be in the +/- 2% range - but calibration drift, etc., can make the error larger. This means that numbers could be off between two riders putting out the same effort by as much as 4% while still being within manufacturers' tolerances - which is large enough to make the difference between winning a GT and being on the podium.

To me the main point is that SRM data is helpful, but not enough to show somebody's guilt or innocence.

Accurate, high-resolution positional data (say by race organization-supplied tamperproof gps trackers) would seem to me to go a longer way toward getting solid reference data for performance.

 

[Race Radio]

Here is a nice video

http://youtu.be/LIRIAinJo0E

Froome attacking everyone. Drops Cobo, Wiggins, Menchov. He set a new record on the climb (17:15) Extraterrestrial. As the commentator says "Wow, wow, wow, wow!"

This year Horner beat Froome's time on the same climb by 33 seconds.

 

[proffate]

As another professional software engineer, I concur that modifying SRM files would be quite easy (or from Brailsford's perspective, inexpensive).

Hell, you don't even have to modify the files, just claim you're a bit heavier than reality. If the p/w doesn't match the VAM models, claim the models are bunk or you had a tailwind or the road surface affected your CRR. If it doesn't match other riders, claim your CDA is better, or their power meters are miscalibrated, or were just inaccurate that day. Remember, power meters are only accurate within +- 2% and are heavily affected by environmental conditions such as temperature, which can change quickly when you're climbing a mountain.

I really don't see how releasing power files is going to prove anything to anyone. Even if you are dumb enough to release files that reveal an FTP of 8w/kg, you can then say the power meter was out of whack. And if a rider is really laying down 8w/kg, then you probably don't need to see his files to raise red flags.

 

[The Hitch]

Here is a nice video

http://youtu.be/LIRIAinJo0E

Froome attacking everyone. Drops Cobo, Wiggins, Menchov. He set a new record on the climb (17:15) Extraterrestrial. As the commentator says "Wow, wow, wow, wow!"

This year Horner beat Froome's time on the same climb by 33 seconds.

Rodriguez and Valverde 2013 also beat Froome 2011 on this climb. By 13 seconds. Yet when they faced Froome at the Tour he outclimbed them no problem.

Similarly Nibali is 8 seconds better than Froome. Yet in the 2012 Tour when they faced eachother he wasn't as good as Froome either.

 

[thehog]

Rodriguez and Valverde 2013 also beat Froome 2011 on this climb. By 13 seconds. Yet when they faced Froome at the Tour he outclimbed them no problem.

Similarly Nibali is 8 seconds better than Froome. Yet in the 2012 Tour when they faced eachother he wasn't as good as Froome either.

Interestingly enough Froome 3 weeks before he was going all cobo on cobo could barely hold onto the peloton.

Which makes you think... how?

Tour de Pologne 2011

Final general classification

1 Peter Sagan (Svk) Liquigas-Cannondale 26:40:00

85 Christopher Froome (GBr) Sky Procycling 0:26:18

 

[thehog]

Anyway I think Prati di Tivo is our best and only indication so far of C v C.

Coming back to Prati Di Tivo the best times are interesting.

Froome took 30 seconds or Nibili's 2012 time and a full minute off Horner's 2012 time!

Horner rode better in 2013 where Froome killed him in a man to man arm wrestle.

Funny that Horner with his broken knee cap rode faster than his 2012 time

-----

2013:14,7 km@7,0%---38:22---average speed 22.99 km/h(Chris Froome)
---38:28---average speed 22.93 km/h (Santambroggio)
---38:33---average speed 22.88 km/h(Vincenzo Nibali)
---38:35---average speed 22.86 km/h(Michal Kwiatkowski)
---38:37---average speed 22.84 km/h(Horner-Contador)
---38:42---average speed 22.79 km/h(Rigoberto Uran)
2012:14,7 km@7,0%---38:56---average speed 22.65 km/h(Vincenzo Nibali)
---39:12---average speed 22.50 km/h(Kreuziger-Horner)

http://climbing-records.blogspot.com/2013/03/froomey-new-best-time-up-prati-di-tivo.html

 

[DirtyWorks]

I'll concede: this is not trivial, but this is not that hard, either - certainly not "epic."

What's an attack? What's the field surging? What's a steep incline? Reviewing each file, every kilometer. What about tracking errors?

The scale and complexity of the work grows quickly as soon as you try to strategically edit gps tracks in some kind of stealthy-fraudulent way.

 

[sittingbison]

...Hell, you don't even have to modify the files, just claim you're a bit heavier than reality...

Horner edited his wikipage weight from 70kg to 63.5kg in October 2009 (thanks ElChignon ) and it looks like he is at least a couple of kilos lighter now than 2009...the recent comments are indicating about 60-62kg.

Dawg is observably much thinner now than he was in 2009...10kg?

Sir Wiggo went from 82kg in 2008 to 69kg in 2011/12 but apparently has gone back up to 80kg or so now.

Yet all are putting out their best ever power at the skinny end. And what effect does a couple of kilos have on the watts/kg factor? What effect does 10kg have? What effect does it have when the total power is equal or more when 10kg less?

Lets see now:

There is some manipulation going with riders weights. That riders are not honest about their weight is not on. If the riders weight is unknown does that not make the maths a guessing game?

 

[ChewbaccaD]

Here is a nice video

http://youtu.be/LIRIAinJo0E

Froome attacking everyone. Drops Cobo, Wiggins, Menchov. He set a new record on the climb (17:15) Extraterrestrial. As the commentator says "Wow, wow, wow, wow!"

This year Horner beat Froome's time on the same climb by 33 seconds.

Yea, but that was 2011 Froomedog. 2013 Froomdog appears to be a much improved animal...and that is why people scratch their head at his meteoric rise.

 

[Scott SoCal]

Yea, but that was 2011 Froomedog. 2013 Froomdog appears to be a much improved animal...and that is why people scratch their head at his meteoric rise.

It's a natural progression given his obvious pedigree.


<face palm>

 

[ulrichw]

What's an attack? What's the field surging? What's a steep incline? Reviewing each file, every kilometer. What about tracking errors?

The scale and complexity of the work grows quickly as soon as you try to strategically edit gps tracks in some kind of stealthy-fraudulent way.

Since this is getting fairly OT, I'll try to keep my answer short. I would simply seed the algorithm with filters based on my rider's known maximal efforts (e.g., peak 5m effort, peak 10m effort, etc.). The algorithm would then simply "tone down" any peaks that are too close to these maximal efforts, making, in effect, my rider seem more "normal". It doesn't matter what generated the peaks, but if you want to be really sophisticated, you could just tell the program to focus on known segments of the race (e.g., the last climb). I'd run this against a few sample files to make sure the parameters for toning down are neither too aggressive or too tame, and then I'd be set. All future files would simply get passed through this program before publishing.

Looking at each file, every kilometer is very easy for a program to do - especially since it can very easily discard the 90-ish percent of the time when the rider is nowhere close to a maximal effort.

 

[Race Radio]

Yea, but that was 2011 Froomedog. 2013 Froomdog appears to be a much improved animal...and that is why people scratch their head at his meteoric rise.

Froome's Meteoric rise came in 2011. There was a huge leap from 2010 to 2011 but since then the majority of his improvement has come in this TT ability and his pacing, which was garbage

 

[Race Radio]

Rodriguez and Valverde 2013 also beat Froome 2011 on this climb. By 13 seconds. Yet when they faced Froome at the Tour he outclimbed them no problem.

Similarly Nibali is 8 seconds better than Froome. Yet in the 2012 Tour when they faced eachother he wasn't as good as Froome either.

Riders are not robots that perform the same every race, they don't. Based on W/kg I have not seen a climbing performance in the last 5 years that equals Horner on the Angrilu.

 

[Fearless Greg Lemond]

The best SRM file, and weight, to have would be Quintana on the Semnoz. Froome did not go full gas that day but still would be interesting to see as the estimates for Quintana are 6-6.3 w/kg

I disagree, Quintana drafted the whole climb behind Rodriquez.

For me it is not a popularity contest, or about who doped more....but the numbers support the idea that Horner climbed faster on multiple climbs in the Vuelta then Froome ever has.

You mentioned Pena Cabarga upthread, do you know Froome's last K was 20 seconds faster in 2011 than Horners?

But please, compare Alto de Hazallanas with Axe 3 Domaines. Or Ventoux with Angliru.

All were great freak shows.

 

[red_flanders]

Interestingly enough Froome 3 weeks before he was going all cobo on cobo could barely hold onto the peloton.

Which makes you think... how?

It makes me think Dauphine style blood draw...

 

[Dear Wiggo]

It makes me think Dauphine style blood draw...

Whilst I had the same thought, it would seem ill prepared, given riders in the past have cycled (!) their blood, maintaining volume whilst refreshing the stored bags. Plus you'd think it would get flagged easily in the BP.

 

[heart_attack_man]

Interestingly enough Froome 3 weeks before he was going all cobo on cobo could barely hold onto the peloton.

Which makes you think... how?

3 weeks... LOL.

Isn't that the usual blood withdrawal timeframe?

ETA: Beaten.