Fighting Confirmation Bias in Research

Confirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms your preexisting beliefs.

You develop a thesis: "XYZ company is undervalued." Then you research. But your research is systematically biased. You notice every analyst who agrees with you. You rationalize away disagreeing analysts. You focus on metrics that support undervaluation while downplaying metrics that suggest overvaluation.

By the time you complete your "thorough analysis," you've actually just conducted a very thorough search for confirmation of a preexisting conclusion.

This is distinct from other biases. It's not about emotion. It's about the very process of information gathering becoming corrupted by what you want to be true.

The result is that analysts with strong initial beliefs often reach stronger conclusions after research, not because their initial belief was validated, but because research is inherently biased toward validation.

Quick Definition: Confirmation bias is the systematic tendency to seek, interpret, and remember information in ways that confirm your preexisting beliefs about a stock or market, while dismissing or underweighting contradictory evidence.

Key Takeaways

• Confirmation bias warps the research process itself: It's not that you're consciously distorting evidence. It's that your brain automatically guides your research toward confirming evidence and away from disconfirming evidence.
• You're most subject to confirmation bias when you have strong initial beliefs: The more certain you are before research begins, the more biased your research will be.
• The antidote is actively searching for disconfirmation: Instead of trying to prove your thesis, try to disprove it. Systematically seek evidence that contradicts your position.
• Outside view helps: Getting perspectives from people who don't share your thesis helps identify evidence you're dismissing too easily.
• Written frameworks reduce confirmation bias: If your thesis is precommitted in writing before research, it's easier to notice when you're rationalizing evidence away.
• Quantitative screens can bypass confirmation bias: Rather than manually researching stocks, using systematic screens removes human judgment (and thus confirmation bias) from position selection.

The Mechanism of Confirmation Bias

Confirmation bias works through three cognitive pathways:

Selection bias in what you read: You want to understand a company. So you read the latest quarterly earnings call. But earnings calls are management presentations—they emphasize positive news. You read sell-side analyst reports. But sell-side analysts are anchored to the most recent price. If the stock has just declined 30%, the analyst revises downward from a thesis that was too bullish, not to a newly accurate level.

You end up reading sources that are systemically biased toward confirming your initial belief (or toward recent price moves).

Interpretation bias in how you read it: An earnings call announces that "costs increased 10% while revenue increased 12%." This is margin compression. An investor wanting to believe the stock is undervalued interprets this as "revenue growth is ahead of cost growth, margins will improve with scale." An investor wanting to believe it's overvalued interprets it as "margin pressures are emerging."

The same data point. Different interpretations based on preexisting belief.

Memory bias in what you remember: A week later, you recall the earnings call. Your memory preferentially retrieves: "Revenue growth 12%" and largely forgets: "Cost pressures are emerging." Your brain automatically remembers the confirming parts more vividly.

Together, these three biases mean that after "thorough research," your original thesis has been reinforced, sometimes quite dramatically, regardless of whether new evidence actually warranted that reinforcement.

How Confirmation Bias Manifests in Investment Research

1. Due diligence becomes myth-building

You decide a company is undervalued. Then you conduct due diligence. But "due diligence" becomes a process of constructing a coherent story about why it's undervalued. You notice management execution risks in the bull case. You ignore management quality risks in the base case.

By the end, you've constructed a myth ("this company will transform under new management") that confirms your original thesis.

2. Bull case gets more detailed than bear case

When researching, most investors develop a detailed bull case and a cursory bear case. The bull case has 10 specific paths to profit. The bear case is "growth could slow, competition could intensify, margins could compress."

This asymmetry isn't accidental. It's confirmation bias. You're building a case for the position, so you naturally develop the bull case more thoroughly.

3. You pattern-match toward confirming patterns

The company had a great year in 2015. You notice this and interpret it as evidence of management quality. The company had a terrible year in 2017. You notice this and interpret it as "cyclical headwind, not management failure."

Same management team. Different years. Your interpretation of the same manager's competence changes based on which year you're emphasizing.

4. You reframe evidence

A metric that would suggest overvaluation if your thesis were wrong gets reframed. "The stock is expensive on P/E, but that's because earnings are temporarily suppressed" (bull case). "The stock is cheap on P/E, but earnings are peak cycle" (bear case). Same metric. Different interpretation.

The Cost of Confirmation Bias

Poor calibration of thesis confidence: You believe your thesis is well-founded because you've "thoroughly researched" it. But your research has been biased toward confirmation. Your actual confidence should be lower.

Vulnerability to unexpected evidence: When the market reveals information that contradicts your thesis, you're unprepared. You've spent so much time constructing the confirming case that you haven't deeply explored the disconfirming case.

Overconcentration in positions: Because your thesis feels rock-solid (thanks to confirmation bias), you take larger positions than warranted. When the thesis breaks, the loss is larger because you were more confident than justified.

Holding positions past invalidation: You're biased toward interpreting new data as confirming rather than disconfirming. So positions that should be sold are held because you're biased toward interpretations that support holding.

How Successful Researchers Counter Confirmation Bias

Buffett's approach: Institutional investor committee

Berkshire's investment committee intentionally includes people with different views. The process forces discussion of bear cases and disconfirming evidence. You can't just present a bull case and have it approved.

Munger's approach: Steelman the opposition

Charlie Munger explicitly argues against his own positions before investing. Not to convince himself to sell, but to surface disconfirming evidence he might otherwise dismiss.

Damodaran's approach: Quantify assumptions

Aswath Damodaran makes every assumption in a valuation explicit and numerical. Then he stress-tests: "If this assumption is wrong by 20%, what's the impact?" This forces honest reckoning with sensitivity rather than confirmation bias.

Klarman's approach: Pre-mortems

Before taking a major position, Klarman's team conducts a "pre-mortem." They assume the position loses money. What went wrong? Why did the thesis break? This forces explicit disconfirmation before commitment.

Practical Techniques to Counter Confirmation Bias

1. The Devil's Advocate Framework

For each position, assign someone to explicitly argue against it. Not to make you sell, but to surface disconfirming evidence that confirmation bias might cause you to dismiss.

The exercise: present the strongest possible bear case. Not the strawman bear case that's easy to refute, but the most credible bear case.

2. The Pre-Mortem Technique

Before investing, assume the position will lose 50%. Write down: "What went wrong? Why did my thesis break?" This forces you to pre-commit to disconfirming scenarios before emotion is involved.

3. Seek Opposing Views First

Instead of researching to confirm your thesis, start by reading the strongest opposing view. Read a bear-case analyst. Read short seller research (with skepticism). Read activist investor demands if applicable.

Get the disconfirming evidence first, before confirmation bias has shaped your initial view.

4. Quantify Your Thesis

Write down your thesis numerically. "I believe revenue growth will be 5-7% over next three years. Margins will expand by 50bps annually. Intrinsic value is $X. Margin of safety is Y%."

When new data arrives, compare it to these explicit predictions. If you find yourself adjusting the predictions rather than updating the thesis, confirmation bias is operating.

5. Build a Decision Journal

Record your thesis before investing. Include explicit disconfirming scenarios: "This position is wrong if X." Then update: "Did X happen? Did I change my opinion when it happened?"

This reveals whether you're exhibiting confirmation bias (reinterpreting disconfirming evidence as irrelevant) or genuinely updating (exiting when thesis breaks).

6. Use Quantitative Screens

Rather than manually researching a stock and becoming attached to a thesis, use quantitative screens to select positions. The screen removes human judgment and thus confirmation bias.

Downside: you lose the ability to apply deep judgment. But you eliminate confirmation bias in the selection process.

7. Diversity of Perspectives

Discuss positions with investors who have different investing styles. A growth investor will notice risks that a value investor overlooks. A macro investor will see market structure issues that a bottom-up investor misses.

Different perspectives expose confirmation bias by challenging your frame.

8. Red Team Your Research

Assign someone to specifically try to poke holes in your analysis. Not in a hostile way, but with genuine curiosity: "Where might this be wrong?"

This is more effective than self-critique because you're less likely to confirm your own biases than to accept external critique.

Real-World Examples

Technology analyst bubble endorsement

During the dot-com boom, sell-side analysts who initiated coverage of tech companies faced confirmation bias. Their research started with: "This is a hot stock, I should recommend it." Then their analysis followed, selectively gathering evidence of why tech would disrupt everything.

Disconfirming evidence (these companies weren't profitable, had no sustainable advantages, traded at 100x sales) was dismissed as "misunderstanding the new economy."

The confirmation bias meant that analyst coverage became a leading indicator of overvaluation, not undervaluation.

Kodak's digital dismissal

Kodak's photography division executives knew that digital photography was coming. But their research and internal analysis was biased toward confirming the thesis that film would remain dominant. Disconfirming evidence was reframed: "Digital will be a niche product," "Consumers prefer film quality," etc.

The confirmation bias prevented Kodak from adapting despite having the information that adaptation was necessary.

Value investor's housing crash thesis

A value investor had a thesis that housing was overvalued in 2006. Then they conducted detailed research on mortgage-backed securities. But the research was biased toward confirming the overvaluation thesis.

When housing didn't collapse immediately, the investor's confirmation bias caused them to hold the position longer and larger than they should have. They were right on the direction (eventually) but early enough that capital was destroyed waiting for validation.

Overconfident analyst on Amazon

An analyst in 1999 had a bearish thesis on Amazon. Then they researched to confirm that thesis. They found plenty of evidence: "E-commerce will never be profitable," "Retail is controlled by Walmart," "No moat in online retail."

All this evidence confirmed the thesis. The analyst was so confident they shorted the stock. Confirmation bias caused them to miss the disconfirming evidence: Amazon's customer acquisition efficiency, its willingness to sacrifice profitability for scale, the emerging infrastructure advantages of being first.

The confirmation bias made them overconfident, and overconfidence led to catastrophic losses.

Common Mistakes

Mistake 1: Believing that "thorough research" eliminates confirmation bias More research often amplifies confirmation bias. You're just gathering more confirming evidence.

Mistake 2: Assuming you're not subject to confirmation bias Almost everyone underestimates how much confirmation bias affects them. The more you believe you're objective, the more likely confirmation bias is operating invisibly.

Mistake 3: Using confirmation bias as an excuse to not research The solution isn't to avoid research. It's to actively search for disconfirmation while researching.

Mistake 4: Holding positions that contradict your thesis because you've "invested too much time" If your thesis breaks, exiting isn't admitting failure. It's responding to new information. The time you've invested is a sunk cost.

Mistake 5: Never re-reading opposing research You read the bear case once. Then confirmation bias causes you to forget the strongest points. Re-read it quarterly. Force yourself to reckon with disconfirming evidence.

FAQ

Q: Is confirmation bias the same as wishful thinking? A: Wishful thinking is conscious ("I hope this works"). Confirmation bias is unconscious—you genuinely believe you're being objective while your brain guides you toward confirming evidence.

Q: Can you eliminate confirmation bias? A: Largely no. But you can reduce it substantially through systematic processes: decision journals, pre-mortems, seeking opposing views, quantifying theses.

Q: Is confirmation bias worse in bull or bear markets? A: It's equally powerful in both. In bull markets, you confirm bullish theses. In bear markets, you confirm bearish theses. The bias is to whatever position you've initially committed to.

Q: Does expertise reduce confirmation bias? A: Somewhat, but not much. Expert investors can fall victim to confirmation bias as much as novices—they're just more credible when they do.

Q: How do you know if your research has been biased? A: Check your conclusion before and after research. If your thesis has gotten dramatically more confident and extreme after research, confirmation bias likely operated. If it's more nuanced and you've genuinely updated, that's better.

Q: Should you ever just trust your thesis without second-guessing? A: Yes, at some point you need to act. But the second-guessing should happen during research, not after you've committed capital.

Q: How do you balance conviction with openness to disconfirmation? A: Have conviction in your process and framework. Have humility about your specific conclusions. The process might suggest a stock is undervalued. But the specific prediction of how much and when might be wrong.

Related Concepts

Selective exposure: The tendency to expose yourself to information that confirms your view. Related to confirmation bias but distinct—this is about what sources you consume.

Bias blind spot: The tendency to believe you're less subject to biases than others. This meta-bias makes confirmation bias more powerful.

Motivated reasoning: The tendency to generate arguments in favor of conclusions you're motivated to believe. This is how confirmation bias operates—through reasoning processes.

Hindsight bias: The tendency to believe past events were more predictable than they were. This complements confirmation bias by making you believe your original thesis was obviously correct.

Summary

Confirmation bias is one of the most insidious distortions in research because it operates invisibly. You believe you're conducting objective analysis while your brain systematically guides you toward confirming evidence and away from disconfirming evidence.

The result is that your thesis becomes more confident after research, but not necessarily more accurate. You've just constructed a more convincing case for a potentially flawed conclusion.

The antidote is systematic: actively search for disconfirming evidence, quantify your thesis so you can update when it breaks, get external perspectives, use pre-mortems to surface risks you're biased toward ignoring.

The best investors haven't eliminated confirmation bias. They've built systems that make it visible and create incentives to correct it. They make space for the devil's advocate. They explicitly ask: "What would prove me wrong?" They force themselves to research the bear case as thoroughly as the bull case.

This isn't comfortable. It means genuinely wrestling with disconfirming evidence rather than dismissing it. But it's the only path to research that's actually objective rather than just carefully constructed confirmation.

Next: The Endowment Effect and Value Traps

Confirmation bias causes you to hold positions despite disconfirming evidence. The endowment effect causes you to overvalue positions simply because you own them.