All eyes on Punxsutawney Phil this morning, and the news from Gobbler's Knob is in:
He saw his shadow. Six more weeks of winter and colder weather...or so the story goes.
The data isn't so clear.
A while back, The Washington Post ran the Groundhog Day numbers for the last 30 years:
Are weeks following shadow-years really colder than non-shadow years?
Are weeks following non-shadow years really warmer than shadow years?
Turns out, not so much. The pattern is precisely what we'd expect to see in a large data set of cities, and a random outcome each year. Sometimes Phil has it, and sometimes he doesn't.
What's notable, and relevant to becoming a better investor, is how terrible Phil's forecasts have been for St. Petersburg, Florida. If you lived there, you'd assume Phil had no skill:
In fact, shadow years – which are supposed to be cooler – were 13.9 degrees warmer than non-shadow years! He's not just wrong, but he's wrong by a large magnitude.
And that's the point. It's an outsized, tremendously incorrect forecast...but it's completely within the realm of expectations for a large data set. If you look at the data points for every American city, as this study did, you're bound to discover some funky outcomes.
In larger doses, random outcomes get streakier, and can allow for more salacious storytelling. But it doesn't mean they're necessarily meaningful. It might just mean you have a larger data set, and more opportunities for results to drift from expectations.
What we know from investment conversations is people don't talk very much about their terrible data points. They talk about their winners. And guess what else exists when there is a large enough data set? Outsized, tremendously correct forecasts.
There is always Oklahoma. Years when Phil predicted cooler weather, Oklahoma City was indeed 8.5 degrees cooler than years when he didn't. If you lived there, you'd think Phil was a whiz with uncanny predictive power.
-
We would all be better off to avoid extrapolating that one person's successful investment story is their only investment story, and that one person is every person.
Large data sets tell us that we should expect crazy stories of both success and failure. Human nature tells us that we will most often hear about the successful ones.
This is the backdrop of investment storytelling.
And isolated anecdotes lead to misinterpreting outcomes...these one-off stories are how luck by the investor gets confused for skill by the listener. If anyone makes a lot of decisions, great outcomes happening randomly are no longer surprising, but expected.
Caveat emptor.
There is always Oklahoma.