Friday, November 23, 2007

Confirmation Bias

My first blog post on May 12, 2005 reviewed Nassim Nicholas Taleb's wonderful book, Fooled by Randomness.

I am currently reading his second, The Black Swan, the Impact of the Highly Improbable.

The first book centers on the "hidden role of chance in life and in the markets." As an investor, it is particularly apropos as one tries to identify systematic methods of creating value via investing. Investors and investees like to believe that the investment world is deterministic with clearly understood cause and effect. Understanding causal drivers of value helps to create repeatable models for investment that scale both across time and individuals in the firm. Nassim challenges us to be very careful in over ascribing reason and logic to an outcome. Too often, investment outcomes are the result of randomness rather than science, ie being lucky rather than good.

The second book posits that we expect the world to operate within very narrow bands of probability and tend to discount the possibility of extreme events, or black swans. For example, Wall St uses Value At Risk models that analyze capital at risk with a 95% confidence interval - the models, based on Monte Carlo simulations, create a range of statistically probable outcomes. The savings and loan debacle, the current subprime mortgage mess, Long Term Capital, etc illustrate the fallacy of discounting the highly unlikely.

The book also introduces a key concept that with real application for venture capital - confirmation bias. He writes, "cognitive scientists have studied our natural tendency to look only for corroboration; they call this vulnerability to the corroboration error the confirmation bias." Due diligence is the process by which investors analyze and evaluate investment opportunities. Too often, however, diligence is confirmatory in nature. As Taleb notes, "...subjects supplied mostly questions for which a "yes" answer would support the hypothesis. Disconfirming instances are far more powerful in establishing truth. Yet we tend to not be aware of this property. A series of corroborative facts is not necessarily evidence. We can get closer to the truth by negative instances, not by verification."

In my experience, investors often look for corroboration at the cost of negative empiricism - ie at the cost of looking for non-confirming evidence. Such evidence in itself may not argue against the deal, it will, however, help avoid Pollyannish projections based more on hope than on truth.


  1. That's a real neat take away, and one that companies could ask about all elements of their "self analysis" in terms of their (our) exposure to risk of unexpected market changes etc.

  2. I agree with everything in this post. ;-)