The Use (and abuse) of History

Humans often think that historical patterns can be used to predict future actions. We think that a meteorically successful businessperson will continue to be so successful. We think that a market that has gone mostly up for decades will not crash. We think that a successful business is invulnerable.

I submit that these expectations should be tempered with an understanding of why the pattern was created, any mechanisms supporting them, and the weakness, limits, or instability of those mechanisms.

A Ponzi scheme looks really good until it collapses. History of continuous value increase is misleading unless you understand how the increase is being borrowed from the future.

I ask you to endure a “thought experiment”. I saw this one in a management volume somewhere, debunking the theory that there exist managers who consistently made decisions without sufficient information, and who had been consistently successful.

The point of the thought experiment is to show that a set of facts about something should not be taken without consideration of how each instance was selected.

Imagine I have a box of pennies, perhaps ten thousand of them. If I shake the box and note which pennies came up heads, and do this again five more times, odds are there are some pennies that have come up heads all six tries. (You can play with different size numbers here, to get an apparently even more unlikely result).

I now point out one of these tail-free pennies and claim that it always comes up heads. I have statistics to prove that it does.

You will refuse to believe that this is a property of this specific penny, if you saw how I selected it. But if you only knew that I had a six-time winner by betting on heads, you might believe the penny was “loaded”.

Beware when people give you statistics on anything. Look for the generating mechanism(s); look for the after-the-fact selections too.

Our stock market, as Didier Sornette pointed out, has never suffered from an invasion, hyperinflation, and many other things that can lead to a crash. So we assumed it could not crash, because it never did. Removal of protections over the years certainly contributed to the debacle. Those protections were put in about 80 years ago – after a crash. Those protections helped generate those nice statistics.

Leave a Reply

Your email address will not be published. Required fields are marked *