Tuesday, July 27, 2010

Art of the Extreme Forecast

Here is an article in the Wall Street Journal by Jason Zweig on why it pays to say something extreme if you are in the prediction business. He starts out mentioning the prediction earlier by Robert Prechter earlier this month that the Dow Jones Industrial Average will drop below 1,000 within six years.

An extreme forecast doesn't merely grab your attention; ironically, it may strike you as even more convincing than a moderate prediction. A classic psychological experiment at the University of Michigan showed that 54% of people preferred an extreme prediction about stock prices to a more-temperate one. They apparently believed that a forecaster must have high confidence and a solid rationale in order to justify making a dramatic prediction. So, while Dow 1000 may or may not be a good forecast, it isn't bad marketing for newsletters...

The more extreme predictions tend to be popular at turning points. Zweig points out that the book "Dow 36,000 hit the best seller list around the year 2000. The article also points out that for Dow 1,000 (a 90% drop) to happen earnings would have to collapse and price to earnings would have to drop to record lows.

One of the examples given is Pfizer. What would it take for it to trade 90% lower? Earnings would have to contract 70% and it would have to sell at a price/earnings of 2.3.

Zweig also points out that if Coca-Cola ever fell to a PE of 5, Warren Buffett would buy the whole company pretty quickly.

Adam

This site does not provide investing recommendations as that comes down to individual circumstances. Instead, it is for generalized informational, educational, and entertainment purposes. Visitors should always do their own research and consult, as needed, with a financial adviser that's familiar with the individual circumstances before making any investment decisions. Bottom line: The opinions found here should never be considered specific individualized investment advice and never a recommendation to buy or sell anything.
 
Site Meter