Kittel sprints to victory!

Yesterday's stage and today's stage provided us with a real learning experience, and a humbling one at that.  Last year was the first time we made adjustments to our model while the Tour de France was taking place.  We tried to make tweaks based on strategies we thought were in play.  For the most part, our adjustments worked extremely well.

Now to today's learning experience.  Marcel Kittel won today's stage 4 with the time given below.

• Stage 4:   5h 28' 30" (actual), 5h 56' 59" (prediction), 28' 29" slow (8.67% error)

After yesterday's enigmatic slow stage, we thought cyclists were holding back, saving energy for later stages.  Given that today's stage was longer than yesterday's stage, we thought power outputs would be reduced again.  As I noted yesterday, we would have needed to cut power by a third to hit yesterday's time.  We decided to cut power by 20% today, thinking we'd see another slow stage.  We should not have done that!  I cringed when I saw that had we not altered our model by reducing the power, we would have missed today's stage by just 47".  Getting spooked by yesterday's stage caused us to think that teams' strategies were such that we had too much power for the long, flat stages.  We should have seen yesterday as an anomaly and trusted our model for today's stage.  But we can't go back in time and claim a great prediction today.  Our model performed great; we misread the cyclists' strategies.  Part of good science is being as honest as possible.

Below is Kittel's average speed.

• Stage 4:  12.05 m/s (43.38 kph or 26.95 mph)

We are going to trust our model for tomorrow's Stage 5, which is a 216-km (134-mi) medium-mountain stage that starts in Limoges, where Stage 4 ended.  Cyclists will head southeast to the mountain resort of Le Lioran.  Two category-2 climbs near the end of the stage should make for great racing.  Our prediction is given below.

• Stage 5:   5h 29' 33" (prediction)

We'll stick with our original prediction and hope racing tomorrow is more like today than yesterday.