Have you ever been watching a sporting event when they cite something like this:
“Over the course of the last 70 years the New York Giants are 10-0 when they have 3 or more sacks in a game”.
Okay. That sounds interesting. It certainly fills the commentary. But is this sort of commentary even remotely useful? In most cases it’s statistically and historically irrelevant.
First, the sample size is outrageously small. No serious statistician would take a sample of 10 events and conclude that there’s a real meaningful deterministic output here.
Second, the history of this data makes it practically irrelevant. The 2014 New York Giants are not the 2011 or 1986 New York Giants.
Third, the game in which this statistic might be cited is totally unique. The previous games could have been against sub-par opponents or in unique environments. Unless you can capture these statistical data points in a vacuum they are far more misleading than we think.
The funny thing is, we see this in the investing world almost every day. People love to cite historical data as if it somehow means something about the future returns. The problem is, most of this is actually useless. The current environment is nothing like past environments. And citing a statistical data set where we have, MAYBE, a dozen relevant business cycles hardly makes the data useful.
Probabilistic thinkers do not rely on irrelevant datasets to make choices about the future. They might play some role, but smart probabilistic thinkers are naturally forward thinking. They asses the world for what it is today and find highly probable outcomes based on those understandings. Relying on historical data might give you some perspective about the future, but be very careful thinking that that alone will lead to positive future outcomes.