Wednesday, July 01, 2009

6 Tour Wins, Probability & Sporting Misconceptions

We have more or less 3 days to go for this year's Tour de France. It is that time again when various websites, blogs, forums and paper publications devote many many words and pages in prediction of who will win the yellow jersey in Paris. The focus is often on a handful of men. The other 200 or so odd racers who participate with them never grab much media attention at all.

However unfair that may sound, there's some interesting things an astute observer can spot here. Most predictions will pick as favorites a band of individuals based solely on good results in the past. These ones then go into the Petri dish for review. Then one or two are picked and we tell ourselves or our friends, "You know what, I think he's so going to win because he can climb well, 'cause he won so and so stage in the mountains in 2003....and he can time trial like a beast because he won so and so stage at this pace in 2006 and 2007."

This sort of thinking to me is a little absurd. In fact, its a fallacy of the human mind. How do you know your theories about a certain individual winning a race are true? The first mistake of this fallacy is that you pick as your favorite already given their performance. You know something about him before making your decision. Then, you make positive attributions about that individual and even extend the praise to his team, his bike, his coach, his muscular makeup, his lung size, his pedaling style, his cadence, the support of his mommy. daddy and wife and possibly everything under the sun you could conjure up.

Extension of positive attributions based on performance : In this example of self-marketing, the message seems to imply that just because Fabian Cancellara won the Swiss Tour, the product offered must be great. Well, what are the odds that he couldn't have done the same on another competitor's setup? Courtesy : SRAM.

The point here is that it isn't very likely that one would make positive attributions about such things if past performance wasn't known. Because with no certainty can you say that your prediction for the Tour winner can time trial faster or 'hang on with the bunch' on the climbs better than say, rider number 9 in that obscure European team over there. Neither can you attribute with certainty that his team fares better than the other, or his coach is greater than the other, just because he won a race. All this can change. Follow on.

From a statistics perspective, every winning cycling performance should have two components to it. A) The racer's true ability. B) Chance. Chance can be any random event. It could be bad weather, mood fluctuations, an untimely crash, good luck (being in the right place at the right time), the presence of race radios, a mechanical failure... almost anything that could influence the race one way or the other.

The principle of 'regression to the mean' says that today if someone gets podium, tomorrow, they could finish somewhere back in the bunch closer to their true ability. By the same principle, someone who's having a slump can outperform his rivals in the next race, driving his performance closer and past his true ability. This is a scientifically validated phenomenon and I urge you to do further reading on it.

But a number of media journalists suffer from not heeding the role of chance. If someone wins once or in streaks, they'll go after him, his winning bike, his family, his coach, his nutritionists, his team mechanic etc and write excellent things about all of them collectively. This is journalism in retrospective - making positive comments about someone or something after a positive event has happened. In the process, the winner's bike, the bike manufacturer, his manager, his coach and his coaching service, his sports drink maker, his sun glass manufacturer...everyone are in for the limelight and big bucks.

There is a great deal of money to be made and lots of sales potential in this exercise. Sadly, the other folks who also raced in the peleton are veiled in absolute obscurity and we may never come to know who and what their backgrounds were.

There is a natural variation and fluctuation to events in our world that could be seen by sports analysts and fans as the true thing when its not. Random events can make one rider unexpectedly lucky on one day than the other. But the natural world has it that this luck will soon subside and an individual's performance will fall down closer to his actual norm.

Not many consider this role of sheer chance, and randomness in race outcomes. Why? Its plainly boring and too erudite, I suppose. There's no story in it, is there? I think in every sporting fortune, we as humans love to hear a good superhero fairytale. We're made that way. We like to make superheros, superhero stories, and idols out of these individuals. We also like to spend a disproportionate amount of time talking about them.

There are also innumerable books written on superstars, their race winning strategies, their mental prowess, the odds they overcame, their nutritional preparations, the amount of detail and attention they devoted to their bikes etc. Some of these books promise similar results if their readers would stick to the rules mentioned in its pages, while failing to discuss or explore an important topic in our natural world, that of the statistical nature of performance.

Having said this, I'm tempted to discuss the role of regression to the mean, and random events in the life of George Hincapie at the Paris Roubaix (where he tried to ride for himself to win). We could do that another day, I suppose. But feel free to take it upon yourself to analyze his past results. Find out what happened, and why he never won the Queen of the Classics to this day. Check out the variability in his performances. He probably rode as hard in every one of those races as the previous one, but something caused him to be in the top 10 one year, while pushing him beyond 50th place in another.

Its time we appreciated the hidden roles of these events before we stick the superhero label on one rider rider, or disapprove another for a poor show. Let's embrace everyone alike who's racing in this year's Tour. Its one of the greatest human endeavors of our time.

* * *


  1. Anonymous7:52 AM

    This article must be banned!

    It crushes the nerds who can remember who won every race, stage etc, and then predict with 'certainty' the results of upcoming races.

    Bike and equipment manufacturers won't be happy either, if we realise that their products don't actually guarantee us winning performances. We try to ignore this, as we are lazy & just try to copy what the best professionals use, often without even questioning if it will fit us.

    In short, thanks for writing the article & keep up the good work.

  2. Podium437:16 PM

    Well said. Shame on the media. Its really been getting me sick of hearing about all the Lance chanting 24/7. Yeah heck, where are the others?! Isn't there like 20 teams or something?

  3. There's some truth to the article, as well exemplified in the latest French championships where relatively unknown riders (or at least not the popular superstars) won. Same in Britain and Canada :)

    That said, I don't think you can accurately say it's just luck that Freire won 3 WCs, or Bettini got WC/Lombardia, or Boonen winning Flanders/PR. Or Lance doing 7 tours. Or Sastre being super consistent in grand tours.

    Truth is, the are riders who are above average in their capacities, even for tremendously talented pro-caliber athletes. On a field sprint, it is ridiculous to say that Cavendish, OBree and Bahamontes (at their peak) and each one will take 333 wins out of 999. Or a climb up the Galibier. Or a 40 km TT.

    This is the reason cycling teams tend to set up their member with the best chance of winning a given prize, be it a stage, classic, jersey, etc.

    You mention Hincapie, and his quest to win a spring classic. There's at least 20 cyclists as talented and more as him, equally as well prepared, with the proper access to gear, coaching, nutritional expertise, etc. And yes, some luck is needed to win, too. But, paraphrasing Hemingway: it's better to be prepared than to be lucky, so when luck comes you can take advantage from it.

    I completely agree that we should embrace all these amazing athletes, and enjoy their struggles and experiences (like Jens not taking what he considered a hollow win over Gárate), savour their small and large victories (one of my favourite memories is Voeckler turning himself inside out to stay inyellow one more day)and sympathize with their suffering.

  4. What you have written is a thought provoking topic. You should hear Jonathan Vaughter's interviews. He has confidence in his team for the Tour, at the same time, he does talk in terms in statistical probabilities. Everyone needs that little percentage of luck that guarantees good performance. I surely don't know how to quantify things like luck and chance but its there. Thanks for another good article. Hoping for an American win at the Tour!!!

  5. Today, someone in the Velonews forums said something of this effect -

    "I find it funny how Armstrong fans put him in a position so that regardless of his performance, he can never lose.

    If he does well, it was to be expected because he's the greatest.

    If he doesn't do well, then here are the reasons why his performance is still impressive-he's 38, he's been out of the sport for four years...blah blah blah.

    There is a HUGE difference between Armstrong fans and cycling fans.

    Cycling fans will watch the Tour regardless of who's racing, whereas Armstrong fans only tune in to see HIM race.

    I'm not an Armstrong fan, I'm a cycling fan and I just want to see a good race.

    But the contrived drama that Armstrong and his fans bring to the table is unnecessary. I do not believe it's good for the sport to have one guy sucking up all the oxygen, and his fans are blatant about their idol worship to the point where the racing becomes a peripheral issue."

  6. Anonymous4:13 PM

    interesting and some truth here... but vastly underestimating most cycling fans and even, dare I say it most cycling media.

    In truth most commentators are always going on a about the ups-and down, the random chances, in weather, crashes, tactics, form etc that effect pro racing.

    Further more, the gist of most sections of your post suggest that all riders are equal and that looking at past form is not a useful guide as they will return to a mean (the implication being a shared mean). Where as obviously past form gives you the mean for the individuals (over months and years, not one off performances).
    This is how and why people with past knowledge of performances rate, and then generally fairly accurately predict likely finishing places. Of course with variations and order precise ranking errors, but no the less a good sense of who will be 'there' and who will not. Were your central premise true then my mum would have as good a chance of predicting the top 10 in Paris as I have.... fancy a bet? lets say 10,000 euros on that? Go on! put your money where your interpretation of statistics and luck in bike racing is! ;-)


Thank you. I read every single comment.