Saturday, February 25, 2012

Apple, the NYT, and the Law of Large Numbers

The NYT worries about Apple's stock price, imagining apparently that the law of large numbers is somehow relevant to making a prediction of its future course.

At times confusing the law of large numbers, which is about the behavior of means of draws from identically distributed random variables, and an imagined law of numbers that are large, such as Apple's market capitalization, at other times confusing stock variables, such as Apple's market capitalization, with flow variables such as the GDP of countries, and, more basically, never indicating the random variable (the price? the change in the price? some other thing?) that the law of large numbers is supposed to apply to, the NYT makes a complete statistical hash of the entire discussion. There is no reason, for example, to think that stock price draws over time for a particular company represent independent, identically distributed random variables, which is what the law of large numbers applies to, yet that is what one must believe for the arguments in the article regarding the stock price to make sense.

What a mess.

Hat tip: Nic Duquette

No comments: