Skip to main content

Confirmation Bias

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,322 words

Confirmation bias is the tendency to seek out, weight, and remember information that supports what one already believes, while underweighting, ignoring, or explaining away information that contradicts it. The term was coined by English psychologist Peter Wason in the 1960s, and the broad phenomenon was synthesized in Raymond Nickerson's influential 1998 review, Confirmation Bias: A Ubiquitous Phenomenon in Many Guises. The core tension for an investor is that markets constantly emit disconfirming evidence — a thesis that "should" work generates as much falsifying data as supporting data — yet the mind's default is to treat the supporting half as signal and the contradicting half as noise. The result is that a position, once taken, tends to become more entrenched as new information arrives, regardless of whether that information is actually favorable.

How it's formed (the mechanism)

Nickerson and the broader literature distinguish three distinct sub-processes, all of which appear in trading:

  • Biased search. People test hypotheses in a one-sided way, asking questions whose likely answers confirm the belief. In Wason's classic 2-4-6 task (1960), participants told that the triple (2,4,6) fit a hidden rule almost always proposed further ascending-by-two triples to confirm their guess, rarely testing triples that could falsify it. The true rule — "any ascending sequence" — was hard to find precisely because participants ran only positive tests. A long trader who only pulls up bullish news and bullish chart timeframes is running the same one-sided search.
  • Biased interpretation. Given identical evidence, people read it in line with prior beliefs. The canonical demonstration is Lord, Ross & Lepper's 1979 Stanford capital-punishment study: subjects on opposite sides of the death-penalty debate read the same mixed studies, rated the study agreeing with them as methodologically superior, and ended up more polarized than before — "biased assimilation." A trader rationalizing a bearish print as "already priced in" while treating a bullish print as confirmation is doing this.
  • Biased recall. People selectively remember confirming instances. Winning examples that fit the thesis are recalled vividly; the disconfirming counts fade, inflating perceived hit-rate after the fact.

These are largely automatic, low-effort System-1 processes; they conserve cognitive load and protect a coherent self-image, which is part of why awareness alone does not switch them off.

How it's used in practice

Confirmation bias is something to guard against, not exploit. Its trading footprints are well-catalogued:

  • Thesis lock-in. After entering a position, a trader interprets each small favorable tick as proof of being right and dismisses larger adverse signals. In an experimental study summarized in the investing literature, when investors were offered a choice between an article supporting a prior investment and one opposing it, they were significantly more likely to read the supportive one — selective information acquisition feeding the bias directly.
  • Underestimated downside. By filtering out negative information, investors under-weight downside risk; research links confirmation bias to distorted risk perception and, through it, to poorer investment decisions.
  • Sell-side and forecast effects. It is not only retail. Research on analyst forecast revisions finds that the market and analysts themselves process revisions asymmetrically in line with prior expectations, consistent with confirmation bias operating among professionals.

The practical countermeasures the literature converges on are structural, not motivational:

  • Consider-the-opposite / pre-mortem. Deliberately list concrete reasons the thesis is wrong, or imagine the position has already failed and explain why. Lord, Lepper & Preston (1984) showed "considering the opposite" measurably reduced biased assimilation where a generic "be unbiased" instruction did not.
  • Define the invalidation up front. Write the specific, falsifiable condition that would end the trade before entering, so exit is governed by a pre-committed rule rather than in-the-moment interpretation.
  • Actively seek the bear case and seek out informed people who disagree; track a written log so recall can be audited against what actually happened.

Adoption, debate & evidence

Confirmation bias is one of the most established findings in cognitive psychology and a staple of behavioral-finance curricula (CFA materials, Investopedia, Britannica, The Decision Lab). Nickerson's 1998 review documents it across scientific, legal, political, and financial domains. As a qualitative description of how investors process information, it is broadly accepted.

Honest caveats:

  • Is it a "bias" or a rational strategy? Klayman & Ha (1987) argued Wason's results reflect a "positive test strategy" — a generally sensible heuristic for hypothesis testing — rather than a motivated preference for confirmation; under Bayesian standards, positive tests can be informationally optimal depending on prior probabilities. More recent decision-theoretic work (e.g. models with information-acquisition costs and time pressure) similarly shows that seeking confirmatory evidence can be a rational allocation of limited attention. So the label "irrational bias" is genuinely contested even where the behavior is real.
  • Effect sizes in markets are hard to clean. Most direct trading evidence is experimental or survey-based; isolating confirmation bias from related effects (overconfidence, the disposition effect, motivated reasoning) in real P&L data is difficult. Treat specific quantitative claims about its market cost with skepticism — robust point estimates are scarce.
  • Debiasing is only partially effective. Awareness alone does little; even trained, intelligent people show it. Structured interventions like consider-the-opposite help but do not eliminate it, and their effectiveness varies by domain (forensic and architectural-decision studies report mixed results).

Strengths & limitations

As a heuristic, positive testing and belief-consistent processing are not pure pathology: they let people maintain coherent models and act decisively under uncertainty, and prior beliefs should legitimately inform interpretation of ambiguous data (that is just Bayesian updating). The failure mode is asymmetry — applying a tougher evidentiary standard to disconfirming evidence than to confirming evidence, so beliefs stop being responsive to reality.

The #1 misuse in trading is letting confirmation bias convert a falsifiable thesis into an unfalsifiable conviction: continuing to add to or hold a losing position because every data point gets reinterpreted as supportive, with no pre-defined condition that could ever prove the thesis wrong. It is especially dangerous when fused with sunk-cost reasoning and the disposition effect (holding losers), and it is a structural ingredient in boom/bust dynamics when a whole crowd interprets ambiguous news through the same prior.

Sources

Dispute flags: the existence of the behavior is well-supported, but whether it is an irrational "bias" or a rational positive-test strategy (Klayman & Ha; cost-of-information models) is genuinely contested. Quantitative estimates of its dollar cost in real markets are scarce and should not be treated as settled.