Trading psychology: four biases to understand

What does trading psychology help you examine?

Trading psychology concerns how thoughts, feelings and judgments enter trading decisions. A useful starting point is to examine how you read evidence: what you notice, how certain you feel, and what you conclude from recent results. That is a different task from finding a winning strategy or deciding which product to trade.

This guide looks at four selected reading risks: loss aversion, overconfidence, confirmation bias, and judgments based on recent results or small samples. They do not cover all of trading psychology. A label cannot diagnose you, establish whether a trade is suitable, or promise a better return.

The Investor.gov bulletin published on 16 June 2014 describes nine investing behaviors from a 2010 Library of Congress review of U.S. research. Its list includes active trading, the disposition effect and attention to past performance. The four topics here are a separate selection, supported by the studies cited below.

Separate the pattern from the evidence

The same action can have different explanations. Selling, holding, changing a forecast or feeling disappointed does not by itself identify a bias. Start with the particular judgment being made.

Loss aversion needs a reference point

In prospect theory, gains and losses are evaluated relative to a reference point, and a loss can carry more weight than a corresponding gain. For a fictional comparison, use your current balance as the baseline: a decrease of 10 units and an increase of 10 units are equal changes in opposite directions. The loss-aversion question is whether the decrease is given more weight than the equal increase. This comparison does not prescribe a choice or measure your own preferences.

Kahneman and Tversky’s 1979 paper developed a descriptive account of choices under risk. A 2020 multinational replication found broad support for the tested choice patterns, with variation between countries and within individuals. Neither establishes that losses always matter exactly twice as much as gains.

Loss aversion is not another name for risk aversion. Preferring a certain outcome to an uncertain alternative is a different comparison. It is also distinct from the disposition effect: the tendency to sell winning investments sooner while retaining losing ones. Wanting to break even, or holding a position that has fallen, is not enough to establish loss aversion. Tax considerations, information and other reasons can also matter; the action alone does not tell you which explanation applies.

Be specific about overconfidence

Confidence can refer to several judgments. Moore and Healy’s 2008 study distinguishes three meanings. They should not be treated as interchangeable descriptions of an optimistic person.

Task difficulty can affect these judgments, and some tasks produce underconfidence. A confident tone or a favorable prediction is therefore insufficient evidence to label someone overconfident.

Barber and Odean’s 2000 study examined common-stock accounts at one U.S. discount broker during the 1990s. Its trading and net-performance findings were consistent with overconfidence models. It was not a randomized test of readers’ psychology, and trading frequency alone is not a diagnosis or a forecast of someone’s future returns.

Look beyond examples that fit

Finding support for a view is not necessarily a mistake. The difficulty is that the same observation may fit several explanations. If you seek only examples that agree with your preferred account, you may never find out what separates it from a competing one.

In Wason’s 1960 rule-discovery experiment, confirming examples alone could leave an incorrect hypothesis unchallenged. That was a small laboratory task, not a study of trading returns. Its relevance here is the question it raises: what observation could show that an explanation is incomplete?

A contrary claim still needs a reliable source and relevant evidence. Giving an unsupported rumor equal weight to a well-supported report is not a remedy for confirmation bias. Likewise, several headlines repeating one announcement are not several independent confirmations. Check where the information originated and whether the apparent alternatives actually address the same claim.

Recent evidence and small samples are different issues

Recency concerns the weight given to recent information. Extrapolation extends an observed pattern into the future. Small-sample judgment concerns how representative a limited set of observations is assumed to be. These issues can overlap, but they are not the same.

Greenwood and Shleifer’s 2014 analysis found that expectations in historical U.S. investor surveys were positively correlated with past stock returns. Those reported expectations differed from model-based expected returns. This is evidence about those survey series, not proof that a particular reader is biased or that a recent trend must reverse.

Tversky and Kahneman’s 1971 paper examined exaggerated confidence in the representativeness and replicability of small samples, using questionnaires about research decisions. It does not supply a minimum number of trades that proves skill.

Recent information may be important: a new report can change the underlying case. Ask whether the information is new, comparable and independent, and how the observations were selected. A run of favorable outcomes does not by itself show that the underlying process improved. A run of unfavorable outcomes does not make a favorable one due.

Keep the reference class in the denominator

A signal can be common when an event occurs without making the event equally common among signals. Bayes’ rule keeps these two conditional probabilities separate.

Here, 80% of reports with an increase are flagged. For a uniformly selected flagged report, the fraction with an increase is instead 80/170 = 8/17, about 47.1%. The original frequency was 100/1,000 = 10%. The two percentages answer different questions. These invented counts do not estimate any real company, event or signal. Ask which reference population and conditional rates support a probability; when those inputs are missing, the example cannot supply them.

Keep the claim, evidence and outcome separate

A short record can put these questions in one place. The following is an invented information-reading exercise, not a real company forecast, a trading recommendation or a tested method of improving returns.

Revisit the original wording and the evidence available at the time. If the basis changed, record that change instead of quietly replacing the claim. The worksheet organizes information; it has not been shown here to eliminate bias, improve performance or establish product suitability.

A trading result and an assessment of evidence are also different questions. One favorable outcome cannot establish a reliable process, while one unfavorable outcome cannot explain every earlier judgment. For the separate arithmetic questions, see the related guides on risk–reward ratios and expected value.

Common questions

Does feeling worried or disappointed mean I have a bias?

No. An emotion or an isolated action does not identify the mechanism behind a decision. This guide describes selected judgment patterns, not a mental-health assessment. It does not diagnose a reader or provide therapy.

Will keeping a trading journal improve my returns?

That claim is not established here. A record can organize a claim, its evidence, uncertainty and later outcome. The studies cited do not demonstrate that this particular worksheet causes better trading returns or recovers losses.

How many results are enough to prove that a method works?

There is no universal count in this guide. The question depends on the claim, how observations were selected, their variability and whether the process changed. More observations do not automatically remove selection problems, and a short run alone does not establish a stable advantage.

Should I ignore recent news to avoid recency bias?

No. Recent, relevant evidence may change the case. Distinguish updating a view because something material changed from extending a recent pattern without examining its basis. Neither continuation nor reversal follows simply from a streak.

Sources and limits

The sources include a 2014 U.S. investor bulletin, historical U.S. account and survey studies, and laboratory research. Their dates, populations and methods limit what can be inferred. The decision record is fictional and is not evidence that a worksheet improves returns.

1. Investor Bulletin: Behavioral Patterns of U.S. Investors
2. Behavioral Patterns and Pitfalls of U.S. Investors
3. Prospect Theory: An Analysis of Decision under Risk
4. Replicating patterns of prospect theory for decision under risk
5. The Trouble With Overconfidence
6. Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors
7. On the Failure to Eliminate Hypotheses in a Conceptual Task
8. Belief in the law of small numbers
9. Expectations of Returns and Expected Returns
10. Conditional Probability, Independence and Bayes’ Theorem: university course notes for the probability relationships. Probability notes checked: 19 October 2026; the report counts above are fictional.