Week 35 Day 1: Confirmation Bias: The Filter That Distorts Your Financial Reality

Imagine you bought Tesla stock. Now every article about Tesla's sales growth confirms your genius. Every article about production problems is 'FUD' (fear, uncertainty, doubt) from short sellers. You have not changed your mind because you never encounter information that challenges your view -- not because it does not exist, but because your brain filters it out.

How confirmation bias operates in investing: (1) Selective information seeking. You Google 'why Bitcoin will reach $100,000' instead of 'arguments against Bitcoin.' You read forums where everyone agrees with your thesis. You follow influencers who share your views. Your 'research' is actually a confirmation exercise. (2) Selective interpretation. An earnings miss for a stock you own means 'the market overreacted, time to buy more.' The same earnings miss for a stock you are short means 'I knew it, the company is failing.' Same data, opposite interpretation, based entirely on your pre-existing position. (3) Selective memory. You remember your winning trades vividly and your losing trades vaguely. Your mental track record is biased positive, which increases overconfidence. (4) Source credibility shifts. An analyst who says your stock will go up is 'smart and insightful.' The same analyst, when they downgrade your stock, is 'wrong and biased.' The analyst did not change -- your assessment of their credibility changed based on whether they agree with you. The antidote: actively seek disconfirming evidence. For every investment thesis, find the three strongest arguments against it. If you cannot steelman the other side, you do not understand the investment well enough to own it.

Confirmation bias was described by Wason (1960, 1968) and is one of the most replicated findings in cognitive psychology. Nickerson (1998) provided a comprehensive review documenting its pervasiveness across domains. In financial markets, confirmation bias contributes to: (1) the information bubble effect (Sunstein, 2001) -- investors in online communities self-select into groups that share their views, creating echo chambers that amplify extreme positions (crypto communities, meme stock forums, gold bug newsletters), (2) analyst herding (Hong, Kubik, and Solomon, 2000) -- analysts issue forecasts that confirm the consensus because deviating from the herd requires processing disconfirming information, (3) overconfidence (Daniel, Hirshleifer, and Subrahmanyam, 1998) -- investors who attribute confirming evidence to their own analytical skill and disconfirming evidence to 'noise' develop increasingly biased self-assessments. The computational mechanism: Bayesian updating requires weighting new evidence by its diagnosticity (likelihood ratio). Under confirmation bias, confirming evidence receives a likelihood ratio > 1 (amplified) and disconfirming evidence receives a likelihood ratio closer to 1 (attenuated). Over time, posterior beliefs diverge from the objectively correct posterior, leading to overconfidence in the biased direction. Lord, Ross, and Lepper (1979) showed that presenting mixed evidence (both confirming and disconfirming) to biased subjects actually INCREASES polarization -- each side discounts the other's evidence and amplifies their own. The investment defense: index investing eliminates the need for individual stock analysis, thereby eliminating the domain in which confirmation bias operates. You do not need to evaluate Tesla if you own all stocks through VTI.