Wine, Weather and My TI-89 Titanium

Finally, a homework assignment I can get excited about. One of the supplementary readings for my Prob Stats class is an article by Orley Ashefelter and company on the "Bordeaux Equation," which appeared in Chance magazine several years after the original study. Specifically, three profs collected quite a bit of data on sale prices and vintage quality of young vs. old wines and then compared them against weather patterns. They than ran regressions against the data to determine what vintages would be spectacular, based solely on weather - no tasting involved. Parker's response - "a Neandrathal way of looking at wine" and he was slammed by several others. Ironically, Ashefelter's research is based on Bruno Prats's observation of how weather affected his juice at Chateau Cos d'Estournel in the St-Estephe region of Bordeaux.

One of the most common complaints was that applying his equation back against his data set couldn't accurately predict the known data. My take on that is that people don't understand that regression forecasting gives you a general idea of where a value will land, but can vary by a standard deviation in either direction.

Article Link

Interesting summarization appears in the NYT on 3/18/1990 (Wine Equation Puts Some Noses Out of Joint), where Ashenfelter puts it all on the line, declaring the 1989 Bordeaux crop will be the greatest of the century. Again, look at the dates; the NYT writes "these wine are barely three months in the cask and have yet to be tasted by critics."

NOTE: If you want to leave a comment, you'll need to register with blogger.com. I got 3 anonymous comments within 5 minutes of posting and I'm tired of it. Make up a a name and an email and you're good to go...