Definition
Selective validation of an existing thesis while discounting evidence that challenges it.
Example
Rates markets in a tightening cycle often pull traders into this bias. Once a trader commits to "policy has peaked," every soft data print becomes a pillar for the view, while labour strength or sticky services inflation gets framed as lagging noise. When terminal expectations reprice upward, adjustment is slow and costly.
Cognitive Driver
The mind defaults to conserving effort. It locks onto the first coherent narrative, treats it as an anchor, and filters incoming data through that frame. Dissonant evidence gets downgraded. Congruent pieces get amplified. The internal model becomes self-reinforcing.
Market Expression
A trader forms a view, scans news for supporting fragments, ignores alternative interpretations, and reshapes ambiguous signals to fit the thesis. Stops widen. Entries get justified by selective data points. Deteriorating price action is reframed as "noise." Reversals are dismissed as "profit-taking."
Trigger Conditions
- Strong directional moves
- Dense macro news cycles
- High prior conviction
- Environments where one dominant narrative crowds out alternatives
- Markets where a single theme (policy divergence, inflation stickiness, AI exuberance) feels intuitively "correct"
Diagnostic Markers
- You keep adding charts that validate only one direction.
- You re-read the same analyst note because it makes the thesis feel safer.
- You shift timeframes to rescue the original story.
- You describe bearish data as "temporary" but treat bullish data as "trend-confirming."
- Your information intake narrows to sources you already trust.
Cost Profile
- Late recognition of regime shifts
- Persistent drawdowns from thesis-defending behaviour
- Reduced adaptability as volatility regimes change
- Skewed risk-reward because losses are held longer than gains
- Dependency on narrative coherence rather than price structure
Differentiation From Adjacent Biases
- Not anchoring: anchoring locks to a number; confirmation bias locks to a narrative.
- Not sunk-cost bias: sunk-cost is about sticking with a position because you've spent resources; confirmation bias defends the worldview that justified the position.
- Not overconfidence: overconfidence inflates belief in skill; confirmation inflates belief in the story.
Corrective Lens
Treat every preferred explanation as provisional. Force incoming data to compete directly with the standing thesis. Demand symmetrical scrutiny: evidence supporting your view must face the same pressure you apply to evidence against it. Use explicit counterfactual prompts in Heron to log what would convince you to exit or flip the idea.