Michael Mauboussin is an investment strategeist at Legg Mason Capital Management and author of The Success Equation. He spoke with reporter Adam Bluestein.

You've pointed out problems with the way we typically study successful businesses. What are we doing wrong?
The premise is, If you want to succeed in the future, you identify companies that were successful in the past and look for attributes you can emulate. But people attempt to extract lessons from what is mostly a random process. Part of that is creeping determinism--once something has been successful, we start to believe it was the only thing that could have happened. We also undersample failure. By extracting attributes from winners only, we miss the companies that chose the same strategies and failed.

How do you pick models to emulate?
There is a skill-luck continuum in business. In manufacturing, for example, there are certain skills and processes that are repeatable and transferable. But in other areas, like in entertainment or with start-ups, there's a higher level of uncertainty. It was impossible to predict that "Gangnam Style" would be so popular. In that world, you want to bet on as many artists as you can to increase your probability of success.

That sounds pessimistic.
There's a lot of work in psychology suggesting that we're a little delusional and too optimistic. As a CEO, you have to be optimistic, but depressed people are actually more accurate in predicting the world. Recognizing alternative outcomes--and the role of luck--keeps your mind open to other possibilities, so you can manage or mitigate them.