Friday, October 24, 2014

Biological Passport Makes (Made?) for Tighter Grand Tours

My last post generalized about the average speed of the peleton in the Tour de France, and stretched (possibly too far) toward speculation on why the Tour just keep getting faster.  Comments suggested comparing running to cycling is not appropriate, because of two factors: (1) technology; (2) training technology (i.e., power meters).  These are valid suggestions, but I'm not persuaded that we've seen startling leaps in either technology or training.  Bikes aren't, for example, 5-10% more efficient than they were in 1992, and the research I've read suggests power meters don't significantly improve performance, on aggregate.

 But that's old news.  This post looks at gaps between Tour GC contenders.  

Why look at that?  What could we gain from it?  

We could get a sense of whether the top competitors are exhibiting an comparative advantage.  Note, this comparative advantage could be from anything that is not universally known or applied in the peleton: anything from equipment to training to, yes, performance enhancing drugs.  

Year-to-Year Differences in Closeness of Contest for Grand Tours
Last post I showed this graph of the Tour: 

This chart suggests a U-shaped trend, with 2008--and the five years when Lance was voluntarily retired--being the most closely contested GC race among the top ten.

Of course, the sample size is small, so in the following graph I add up the gaps from all three Grand Tours (Giro, Tour, and Vuelta):
The blue area on the bottom is the gap between 1st and 2nd place finishers; the dark orange area above that shows the gap between the 2nd and 3rd place finishers; and so on.

Added up, as you can see from the height of each stacked column, Grand Tours were tightest in 2008; the biggest gaps were in 2006.  There is definitely a 5-year period from 2007 through 2011, of closer GC contention.  It's a U-shaped graph.

Differences Between Grand Tours
The following graph shows the average margin between top ten GC finishers (including those disqualified) at each Grand Tour.
Again, the pattern is U-shaped.  If you're wondering about the 2006 Giro, by the way, that's the contest that [a very doped] Basso won by 9 minutes.

Generally, the gaps are smallest in the Vuelta and largest in the Tour.
I had anticipated greater margins in the Giro and Vuelta.  That's because I assumed both have more climbing, and that climbing would be a significant factor in creating separation.  Both assumptions--(1) that climbing is the most important creator of separation and (2) that the Giro and Vuelta have more climbing than the Tour--may be wrong.  There clearly are a number of other environmental factors; the question is whether it is reasonable to hold those constant and assume any differences are the result of performance.

Conclusion
I'm not sure of the answer.  In fact, the analysis raises more questions than it answers:

  • Why might the Tour, for whatever reason, be the least likely to conclude with tight finishes between GC contenders?  One would think that the most prestigious GT would create courses that create tight GC finishing times.
  • Are the decreasing gaps between GC contenders caused by the biological passport? 
  • Are gaps between leaders in nearly universally doped peleton (i.e., 1992-2005) the same as gaps between leaders in a nearly universally clean peleton?
  • Does doping in a mostly clean peleton confer a larger comparative advantage?  

It'd be interesting to see whether these gaps in finishing are correlated with the courses; for instance, do mountains and time trials truly separate GC contenders from each other?

Lots of questions.

Still, there may be a few answers here.

Foremost, it seems a change occurred sometime between 2005 and 2008. Competition among GC contenders tightened and speeds seemed to fall.

That trend has gradually reversed, and gaps between GC contenders are once again opening up.

This could mean nothing more than this: that some riders are taking advantage of comparative advances in technology or training.

It could mean, also, that some riders are using drugs.

In my next post, I plan to look combining an analysis of average speed and gaps between GC finishers.  Are there common patterns and correlation between the two?

I also intend to tackle the issue of weather and parcours, and how aggregate analysis from all three GTs should mitigate the effect of a particularly rainy or mountainous or time trial-heavy race.