Cognitive Biases in Investing: What Has Actually Been Measured
5 steps · one page
In short
The popular version of behavioural finance is a list. Thirty or forty named biases, each with a definition and a cautionary example, presented as a taxonomy of ways to be wrong.
Scope, and this article is built around a distinction most treatments skip. Some behavioural findings come from the trading records of large numbers of real investors. Others come from laboratory experiments with student subjects deciding about hypothetical sums. Both are legitimate research; they support very different claims. This article says which is which, and where the evidence is laboratory-only it says the transfer to investor behaviour is a separate empirical question rather than an established fact. It supplies no bias checklist, no self-assessment, and no method for overcoming anything — because the documented finding across this literature is that awareness of a bias is largely insufficient to remove it. Findings verified 17 August 2026.
A list of thirty biases is not knowledge, for two reasons. It says nothing about which effects have been observed in people actually managing money as opposed to undergraduates answering questionnaires. And it says nothing about size — an effect that is real and tiny and an effect that is real and large appear identically on a list.
The strongest evidence: the disposition effect
The best-documented behavioural pattern in real investor behaviour is the tendency to sell winners and hold losers.
Odean (1998), in the Journal of Finance, examined trading records for 10,000 accounts at a discount brokerage covering 1987 to 1993. The measure compares the proportion of gains realised against the proportion of losses realised, given the opportunities to do each.
| Finding | Result |
|---|---|
| Relative propensity to sell | Investors were about 50% more likely to sell a winning holding than a losing one, relative to their opportunities to do so |
| Explained by rebalancing? | No |
| Explained by the higher trading costs of low-priced shares? | No |
| Justified by what the holdings did next? | No — subsequent performance did not vindicate the choices |
| Effect on taxable accounts | Sub-optimal; lowered after-tax returns |
| Timing | Tax-motivated selling concentrated in December |
Worked example
Why this finding carries more weight than almost anything else in the field. It has been reproduced in the United States, Israel, Finland, China and Sweden by different researchers using different datasets, and found outside equities entirely — in residential property and in options. Later work replicated it on 30,512 accounts in China and on Australian retail brokerage data. So this is not one study, and it is not a laboratory artefact: it is a pattern in the recorded behaviour of hundreds of thousands of people spending their own money across several decades, several cultures and several asset classes. The original author himself noted the limitation that his data came from discount brokerage accounts and invited replication on other populations — which is what subsequently happened, and what the finding survived.
A second real-behaviour finding, and a route between them
Investors who trade most actively earn the lowest returns on average, and the effect is not explained away by the obvious alternatives. Barber and Odean found that the shares individual investors bought subsequently underperformed the ones they sold — not merely by too little to cover costs, but underperformed outright — and that this held where trading was not attributable to liquidity needs, tax-loss selling, rebalancing or a shift to lower-risk holdings. The arithmetic of what that costs belongs to Bid, Ask and the Spread and the behavioural driver to this pillar's article on overtrading.
The finding that connects the two, and it is the most interesting result in this article. Work by Gödker, Odean and Smeets argues that the disposition effect can generate overconfidence. The mechanism is a learning failure rather than a personality trait. Realised gains are salient and memorable in a way unrealised losses are not — an investor who systematically sells winners and holds losers therefore accumulates a memory of decisions that worked, while the ones that did not remain open, unrealised and comparatively unmemorable. Someone assessing their own ability from that record will overestimate it, and will do so through an honest reading of their own experience. Which matters for how this pillar should be read: the biases are not a list of independent errors to be checked off. At least one of them manufactures the evidence that sustains another. The authors are explicit that the tendency is heterogeneous and does not apply to all investors.
Where the evidence is weaker, and why that matters
Many named biases in the popular literature were established in laboratory settings — hypothetical choices, small stakes, student or convenience samples, single sessions.
That is not a criticism of the research, which was designed to isolate mechanisms rather than to measure market behaviour. It is a caution about the inference: whether an effect demonstrated with hypothetical sums also governs someone deciding about their savings is a separate empirical question, and for many biases it has not been settled.
Two consequences follow for a reader. A bias described without its evidential provenance cannot be weighed — the reader cannot tell whether they are being told about a robust pattern in investor records or an interesting laboratory result. And effect sizes are almost never quoted in popular accounts, which is how a small effect and a large one come to be presented as equally important.
Worked example
What this article does not claim. It does not claim that awareness helps. The consistent finding across this literature is that knowing about a bias does little to remove it, which is why nothing here is framed as a corrective. An article that described the disposition effect and then advised the reader to sell losers would be supplying a trading rule, which this portal does not do, and would also be claiming an efficacy the evidence does not support. Understanding why a pattern exists and being immune to it are different achievements, and only the first is available from reading.
Frequently asked
8 questions
Why is a list of biases not enough?
Because it says nothing about which effects have been observed in real investor behaviour rather than in laboratory conditions, and nothing about size — a real and tiny effect and a real and large one appear identically on a list.
What is the disposition effect?
The tendency to sell winning holdings and hold losing ones. Odean's 1998 study of 10,000 discount brokerage accounts covering 1987 to 1993 found investors about 50% more likely to sell a winner than a loser, relative to their opportunities to do each.
Could it be rebalancing or costs?
Neither explains it, and subsequent performance did not vindicate the choices. In taxable accounts it lowered after-tax returns, with tax-motivated selling concentrated in December.
How well established is it?
Unusually well. It has been reproduced in the United States, Israel, Finland, China and Sweden by different researchers on different datasets, and found outside equities in residential property and options — hundreds of thousands of people spending their own money across several decades, cultures and asset classes.
Do biases interact?
At least two do. Realised gains are salient and memorable while unrealised losses are not, so an investor who sells winners and holds losers accumulates a memory of decisions that worked — and assessing their ability from that record will overestimate it, through an honest reading of their own experience. One bias manufactures the evidence that sustains another.
Is the laboratory evidence worthless?
No — it was designed to isolate mechanisms rather than to measure market behaviour. The caution is about the inference: whether an effect shown with hypothetical sums also governs someone deciding about their savings is a separate question, and for many biases it has not been settled.
Does knowing about a bias help?
The consistent finding is that it does little to remove it. Understanding why a pattern exists and being immune to it are different achievements, and only the first is available from reading — which is why nothing here is framed as a corrective.
Does MarketClue tell me if I am being biased?
No. It publishes no bias questionnaire, behavioural profile or self-assessment, and does not compare anyone's conduct against these patterns. That would be interpreting a person's decisions for them rather than providing information about markets.
References
- Odean (1998) — Are Investors Reluctant to Realize Their Losses?, Journal of Finance 53(5), 1775–1798 —
- Barber and Odean (2000) — Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors, Journal of Finance 55(2) —
- Barber and Odean (1999) — The Courage of Misguided Convictions, Financial Analysts Journal 55(6) (bought shares underperform sold shares) —
- Gödker, Odean and Smeets — Disposed to Be Overconfident (working paper; the disposition-effect-to-overconfidence mechanism) —
- Genesove and Mayer (2001) — Loss Aversion and Seller Behavior: Evidence from the Housing Market, Quarterly Journal of Economics 116(4) (the effect outside equities) —
Educational and informational only — not investment advice, a recommendation, or an offer to buy or sell any security. Investing involves risk, including the possible loss of principal. Worked examples use fictional companies and figures.