Probability neglect: ignoring the likelihood of low-probability extreme outcomes

An investor who holds tail-hedge positions through years of normal markets while paying steady premium is making an implicit probability calculation: the cost of the hedge is acceptable given the probability and magnitude of the tail event being hedged. The same investor who refuses to allocate to equities at all because of fear of a 1929-style crash is making a different implicit calculation. Probability neglect is the documented tendency to make these calculations badly—to focus on the salience of the outcome and ignore its probability, in either direction.

What probability neglect is

Probability neglect was studied formally by Cass Sunstein (2003) in Probability Neglect: Emotions, Worst Cases, and Law. The bias is the tendency to focus on the magnitude of an outcome (typically a vivid extreme outcome) and to discount the probability of that outcome occurring. Investors who exhibit probability neglect treat low-probability extreme outcomes as either much more likely than they are (the over-fear pattern) or as essentially impossible (the under-fear pattern), depending on what is salient at the moment of decision.

The bias is closely related to availability bias and base rate neglect, but it has distinct mechanisms. Availability bias is about which outcomes come to mind; probability neglect is about how the probabilities of those outcomes are weighted once they have come to mind. Base rate neglect is about ignoring statistical baselines; probability neglect is about under- or over-weighting the probability of the specific case once the baseline is known.

The bias operates particularly strongly in contexts where the outcome is emotionally vivid. Tail events in investing—market crashes, fund collapses, single-stock zeros—are vivid enough that the probability dimension is easily lost. Investors who recently saw a vivid tail event over-weight its probability for years afterward; investors who have not seen one in a long time under-weight its probability until it occurs.

How it manifests in investing

The over-fear expression is most visible after major market drawdowns. Investors who lived through 2008 typically maintain elevated cash allocations and reduced equity allocations for years afterward, at meaningful opportunity cost. The behaviour is not based on a recalculated probability of the next 2008—it is based on the salience of the recent one. The investor's risk aversion is calibrated to the available memory rather than to the underlying probability distribution.

The under-fear expression is most visible after extended bull markets. Investors who entered the market in 2020 or 2021 typically had no personal experience of meaningful drawdowns and accordingly took risk allocations that did not survive the 2022 cycle's modest stress. The behaviour was not based on a recalculated probability of a drawdown—it was based on the absence of any salient drawdown memory. The investor's risk tolerance was calibrated to the available experience rather than to the underlying distribution.

Both directions produce systematic mis-calibration of position sizes. The over-fearful investor under-allocates to risky assets and accumulates a smaller portfolio over the long run; the under-fearful investor over-allocates and faces capital loss when the eventual stress event arrives. Both are forms of the same underlying bias—substituting outcome salience for probability calibration.

A third expression is in tail-hedge decisions. The investor who buys put options after a market drop is paying inflated implied volatility for protection that is most valuable when implied volatility is low—exactly the wrong time to buy. The investor who refuses to consider tail hedges in calm markets is making the converse error: declining cheap protection because the salient outcome is the calm one, not the tail one.

The cost

The cost of probability neglect is the gap between optimal allocation under correctly-weighted probabilities and the actual allocation under outcome-salience-driven weighting. The gap is bidirectional: over-fearful investors lose return through under-allocation; under-fearful investors lose capital through over-allocation. Either direction can produce material long-run damage.

For tail-risk specifically, the cost is most measurable in the cumulative wealth gap between investors who maintained appropriate equity exposure through 2008 versus those who exited at the bottom and re-entered late. The gap is typically 30–50% of pre-crisis wealth, depending on the specific exit and re-entry timing—a permanent loss caused by probability neglect at the moment of decision.

The cost is also concentrated in specific decisions made under time pressure. The decision to hedge or not to hedge ahead of a defined event (an election, a central bank meeting, an earnings announcement) is often dominated by the salience of the worst-case outcome rather than its probability. The same investor who would not pay for hedging in calm conditions pays double the price when the event is imminent and the outcome is vivid.

What helps

The structural remedy is to make the probability dimension explicit and quantitative. A portfolio with a defined value-at-risk and expected-shortfall calculation against the historical return distribution provides the quantitative anchor against which intuitive probability assessments can be checked. A 95% expected shortfall of −12% means that, conditional on a 5% tail event, the average loss is 12%—a specific quantification that intuitive probability neglect cannot easily distort.

For position-sizing decisions, the practical remedy is to use long-run historical statistics rather than the salient recent experience. The drawdown profile of the past 50 years includes both calm and stressed regimes; the position size that would have been appropriate across that range is more defensible than the position size implied by the most recent regime alone. Systematic strategies that pre-specify their position-sizing rules avoid probability neglect by removing the discretionary moment at which the bias would otherwise operate.

Probability neglect in pfolio

pfolio's analytics report tail-risk metrics including expected shortfall, value at risk, and maximum drawdown for any portfolio configuration. These statistics put low-probability extreme outcomes into quantitative form, replacing intuition (which typically over- or under-weights extreme scenarios) with the historical distribution of realised tail outcomes.

Related articles

• Base rate neglect in investing: ignoring statistical probabilities in favour of specific stories
• Availability bias in investing: why vivid recent events distort portfolio decisions
• Value at risk (VaR) explained: understanding tail risk in your portfolio
• Expected shortfall (CVaR): a more complete measure of tail risk than VaR