12:25 am - Sat, Jul 26, 2014
3 notes

Quick point on performance monitoring as a formal anti-doping method

the utility is not to create a second performance passport parallel to the bio-passport for sanctioning or to flag riders for more testing.

the point is to improve the specificity and sensitivity of the current bio-passport by integrating it into the biopassport

this could be done through a combined metric such as a power-OFF score 


by using it as a prior (like altitude and gender) to allow tighter thresholds.

the reason you need to do this is that doping is not the only thing that changes blood parameters.

blood parameters are affected by hydration, training load, and illness. as such, the thresholds have to be set wide enough so that these other physiological processes don’t trip the passport. 

for example a retic dropping from 1.0 to 0.6 in the middle of a GT could be a couple units of blood transfusion or a viral suppression. 

since one would make the performance go up while the other down, power could be used to adjust the thresholds accordingly so that the under-performing rider is not flagged while the other matching their power duration profile.

one other quick point is that like blood, power is an indirect measure of doping, training, and health status. indirect markers will fundamentally have the same major pros and cons that all indirect markers do. arguing for one and against the other is a debate strategy best left to politicians. if you want to argue against power, please be consistent and argue against blood as well.

11:28 pm - Fri, Jul 25, 2014
3 notes

wears aero helmet


8:38 pm
1 note
8:18 pm
1 note

nice analysis and gets into the VO2 demands of the performances we saw.

the one note I would make is that it needs to be considered at what altitude Peraud’s VO2 max was measured. If it was measured at sea level then the full 85 ml/min/kg will not be available so the actual percent of available VO2max held will be greater than 85%.

6:27 pm
Q: what are your thoughts on 0 float, fixed cleats?

may work for some but likely to hurt most

6:26 pm
Q: The biggest drawback I'm seeing with your analysis is my 'most likely to have blood that glows' list is mainly populated by key domestics who swing off the front mid-climb. Is there anyway to meaningfully compare their performances? Thanks.

I don’t think we could get enough reliable information on guys who swing off prior to the finish

12:46 pm
5 notes

Descriptive analysis of the 2014 Tour de France with 2013 Comparisons

The final 2014 Tour de France podium isn’t quite settled but its clear that Nibali will go on to win and that Pinot, Peraud, or Valverde will round out steps 2 and 3.

In estimated W/kg normalized to a standard weight, CdA, draft and altitude the power range for the podium was from 6.2 - 6.3 naW/kg for shorter 17 minute efforts and down to 5.6 - 5.8 naW/kg for longer 50 minute efforts.

Nibali was unchallenged in any meaningful way. One of the questions though is whether this was due to Froome and Contador crashing, Quintana not contesting, and Rodriguez still recovering, or was Nibali superman. .

Pulling in last years data things get a bit busy.

Once color coded by year it becomes clear that other than Quintana and Rodriguez poor AX3 performances the 2013 podium was very likely superior to 2014.

Comparing just Nibali to 2013 however, it is clear that his performance was very likely on par with the best from last year.

Head to head with Froome, the level is effectively the same and we may have missed out a hell of a race.

In terms of whether or not the performances are concerning, it is a mixed picture.

No it is not concerning that Nibali outclassed the 2014 contenders. They are simply not on the level that we have seen from other top riders recently.

On the other hand, Nibali, like the 2013 podium, would likely have been competitive when compared to the 2002-2007 podium baseline as illustrated by the DpVAM analysis. The remainder of this years podium would not.

The question then comes down to progress. In 6-7 years, have riders been able to make performances gains that were once universally held as unfathomable from a clean rider?

I have my bias.


I am sure you have yours.

12:54 am
5 notes
a hell of a race 
that almost was

a hell of a race 

that almost was

12:48 am
1 note
Q: In your W/kg plots, stage 14 looks doped and stage 17 looks normal. But Lauren Ten Dam was near the leaders on both stages and has posted power on Strava: 374W for 32 minutes on stage 14 (lost 43s to Nibali) and 373W for 31 minutes on stage 17 (lost 64s). This indicates that something is wrong with your W/kg calculations. My guess is you aren't properly accounting for drafting behind teammates on the lower slopes. If you remove stage 14 from your plots, the 2014 tour looks cleaner. Thoughts?

thanks for the question.

my first pass through is to work in W/kg normalized to a typical CdA, weight, and drafting that i know works fairly well  because over several climbs the results will regress toward a true mean. 

see: http://veloclinic.com/intro-to-power-estimate-modelling/

although this many not give the true W/kg on any specific climb for any specific rider it gives me a power duration relationship that can be then compared across riders and years.

on second pass, if enough power data is available, it is possible to try to go back and anchor the data to the directly measured power. doing so off of one riders power meter data is often not a good idea.

see: http://veloclinic.tumblr.com/post/92487826308/5-w-kg-is-the-new-normal-get-over-it

if there is not enough data available, then we can look at the highest and lowest outliers and check for unusually strong wind conditions. often people tend to screw this up however because they forget that with switch back climbs the riders actually spend more time going side to side then travelling toward the finish line. what this means is that even in a perfect “tail” wind the riders are only getting about a third of the benefit of the wind. and if the wind is a perfect “cross” then this actually slows the riders down as a head wind creates mores drag than the reduction from a tail wind since work against wind is wind velocity squared times velocity times CdA. so when people use tiny wind speeds in examples of how much error is possible, they maybe inflating the issue by 300 ish %. 

lastly in terms of drafting, going from no draft to perfect single file is a reduction of about 25% CdA. so aside from instances of a solo climb or getting towed all the way the differences tend to be low enough that over a couple climbs things average out. 

but to answer your question directly

on the Risoul the 95% confidence interval is 6.15-6.35, the fully wind and draft adjusted estimate by Fred Portelau comes in at 6.16 W/kg. even if Nibali’s time is adjusted down to account for the draft and wind effect he is still well above DpVAM. still a damn fast climb.

10:31 am - Thu, Jul 24, 2014
1 note

Preliminary Comparison 2014 vs 2013 Tour de France through stage 18

Install Headline