If there's one thing I've consistently argued on this blog, it's that predictions are usually a waste of time and money. Instead, test your assumptions. Don't just “make assumptions explicit.” Look for how you might be wrong, because then you can do something about it.
So how did that play out, the morning after the US Presidential election? Leave aside your horror or elation. This isn't a partisan point. No matter what your politics or feelings about the result, there's a pattern of bad decisions and misjudgment here. And everyone will also forget that pattern within weeks.
With hours to go before Americans vote, Democrat Hillary Clinton has about a 90 percent chance of defeating Republican Donald Trump in the race for the White House, according to the final Reuters/Ipsos States of the Nation project.
The Huffington Post put Clinton's chances at 98%. (98%!)
The HuffPost presidential forecast model gives Democrat Hillary Clinton a 98.2 percent chance of winning the presidency. Republican Donald Trump has essentially no path to an Electoral College victory.
Huffpo also rather sneeringly attacked Nate Silver's 538 for estimating Clinton's chances at a mere 65%.
While I love following the prediction markets for this year’s election, the most popular and widely quoted website out there, fivethirtyeight.com, has something tragically wrong with its presidential prediction model. With the same information, 538 is currently predicting a 65 percent chance of a Clinton victory
As for The NY Times, their final prediction was
“Hillary Clinton has an 85% chance to win”
It's easy to criticize in hindsight. But why do people keep doing this? Why do naive people keep believing this kind of faux-technocratic nonsense? It just leads people to damaging self-delusion, not just in politics but in business and markets.
Elaborate models and data are no defense against wishful thinking. “Big data” does not protect you against many kinds of error. Monte Carlo simulations can be foolish. How could people possibly put a 98% chance on an election that was close to the margin of error in the polls, especially after the lessons of the shock results of Brexit, the Greek referendum and many others?
But they did. Financial markets were bamboozled, for example. Again.
Reuters: Wall Street Elite stunned by Trump triumph.
We need a better way to do this. Instead of models, you need an antimodel, which is what I am developing.