Confirmation bias
What is Confirmation bias?
Confirmation bias is the tendency to seek, interpret, remember, or weight information in ways that support an existing belief.
Research systems should timestamp the original thesis, expected catalysts, disconfirming evidence, and invalidation conditions. Search and recommendation features should not learn only from material the user already approves. Deliberately surfacing credible counterarguments is valuable, but false balance should not give weak claims equal evidential weight.
Team governance should also track whose challenges changed decisions and whether hierarchy, incentives, or ownership of the original idea makes contrary evidence harder to raise.
How it appears
An investor may follow supportive analysts, dismiss adverse news as temporary, ask questions that invite agreement, or remember successful predictions more readily than failures. Search algorithms and social groups can reinforce this selection. Strong conviction is not itself bias; the concern is an asymmetric evidential standard for favorable and unfavorable information.
Portfolio consequences
Confirmation bias can prolong weak holdings, increase concentration, encourage averaging down without new evidence, and prevent a thesis from adapting. It can also lead an investor to reject a sound manager after selectively collecting criticism. Teams are vulnerable when hierarchy, incentives, or shared training suppress dissent and turn consensus into apparent independent confirmation.
Example
A manager expects a company's margins to expand. Strong quarterly sales are treated as validation, while rising customer-acquisition cost is dismissed as irrelevant. If the original thesis depended on profitable growth, applying different standards to the two facts is confirmation bias. A prewritten scorecard would force both revenue and unit economics into the review.
Decision controls
Write the thesis, key assumptions, disconfirming indicators, and sell conditions before investing. Assign a devil's advocate or independent reviewer, seek primary evidence, use base rates, and conduct a pre-mortem. These controls should improve challenge without creating permanent indecision or rewarding contrarian arguments that lack quality evidence.
Practical review
Ask what evidence would change the view, whether the same source would be trusted if it supported the opposite conclusion, and whether alternative explanations were tested. Preserve rejected evidence and decision dates. Evaluate accuracy over many decisions rather than one outcome. A contrary price move is information but does not by itself prove the thesis wrong.
Sources and further reading
- Behavioral Patterns of U.S. Investors, U.S. Securities and Exchange Commission