Loss aversion is the documented tendency to weigh losses roughly twice as heavily as equivalent gains, and recency bias is the tendency to treat the latest market stretch as most predictive of the future. In the experiments underlying cumulative prospect theory, Tversky and Kahneman (1992) estimated a median loss-aversion coefficient of about 2.25.
Horison publishes information, not investment advice, and what any behavioral finding implies for a particular portfolio depends on individual circumstances this publication cannot know. This article presents documented research on two biases; it diagnoses no reader and recommends no course of action.
What is loss aversion?
Loss aversion is the asymmetry in how gains and losses are experienced, first formalized in Kahneman and Tversky's prospect theory (Econometrica, 1979). The theory's value function is roughly twice as steep in the region of losses as in the region of gains, which means the distress of a given loss exceeds the satisfaction of an equal-sized gain.
The follow-up experiments in Tversky and Kahneman's 1992 cumulative prospect theory paper estimated the median coefficient at about 2.25: a loss of $100 weighed on decisions roughly like a foregone gain of $225. The estimate is a central tendency across subjects rather than a constant for every person, but its direction has replicated widely across the experimental literature.
What is recency bias?
Recency bias is the overweighting of the newest information when forecasting. In investing it appears as extrapolation: recent returns are treated as the best guide to the next stretch, so long bull markets breed confidence and long drawdowns breed dread, beyond what the underlying data supports.
The evidence is systematic rather than anecdotal. Greenwood and Shleifer's study of investor expectation surveys, published in the Review of Financial Studies (2014), found that investors' expected returns tracked the market's recent past returns, and that those surveyed expectations did not align with subsequently realized returns. Expectations, in short, looked backward while prices moved on without them.
Where do the biases appear in portfolio behavior?
Each bias has a documented portfolio-level signature. Loss aversion contributes to the disposition effect, named by Shefrin and Statman in the Journal of Finance (1985): the tendency to sell winning positions too readily while holding losing ones, because realizing a loss forces the pain to be booked. Recency bias contributes to performance chasing, the pattern of money arriving after strong stretches and leaving after weak ones.
| Bias | Core documented finding | Portfolio signature |
|---|---|---|
| Loss aversion | Losses weigh about twice as much as equal gains (coefficient near 2.25, 1992) | Disposition effect: holding losers, selling winners early |
| Recency bias | Expected returns track recent past returns, not subsequent ones (2014) | Performance chasing: inflows after rallies, outflows after losses |
The two interact. Recency bias supplies the forecast — this stretch will continue — and loss aversion sets the intensity of the reaction once a position moves underwater. Together they produce the documented timing pattern in which the average investor's cash flows lag the market's better periods.
What does the measured behavior gap show?
Morningstar's annual "Mind the Gap" analysis compares dollar-weighted investor returns with the time-weighted returns of the funds those investors hold. The 2024 edition reported that investors captured about 1.1 percentage points less per year than their funds' total returns over the ten years through December 2023.
The size of the measured gap varies with period and methodology, and academic critiques have argued over its magnitude. Its direction, however, repeats: when investors time their own flows, the average result has trailed the funds' printed returns, which is the pattern the two biases predict.
What documented practices address these biases?
Several approaches recur in the literature and in practice, each operating by removing the decision at the moment of maximum bias rather than by sharpening willpower.
- Automatic schedules: fixed contribution dates remove the timing decision that recency bias contaminates, the mechanism behind dollar-cost averaging's documented behavioral appeal.
- Written policy statements: rules set in calm conditions, such as target allocations and rebalancing bands, pre-commit decisions before losses arrive.
- Default enrollment and automation: plan design that makes contribution the default has raised participation rates in documented retirement-plan studies, precisely because it bypasses repeated active choice.
- Recording decisions in advance: writing down the reasoning for each position at purchase creates a reference point for later review that is harder to rewrite quietly.
Each practice has limits. Rules bind only as long as they are followed, automation removes some decisions rather than all of them, and no process makes a loss painless — the biases describe reactions that remain part of the machinery. The documented claim is narrower: structure reduces the frequency of biased choices, which is a statement about process, not about any outcome.
Why do these findings matter for reading performance data?
The biases explain a recurring pattern in how published returns are consumed. A fund's reported return is time-weighted and assumes a single holding period, while the investor's realized result depends on when cash moved — the distinction the behavior gap measures. Reading a performance table with both biases in view means asking not only what the fund returned, but when money actually arrived and left.
That reading is descriptive, not prescriptive. What to do about documented biases is a decision that belongs to each investor's circumstances, ideally weighed with sources this article has cited rather than with the most recent market stretch.
For more context, read What Sequence-of-Returns Risk Means for Retirees.
For more context, read What Dollar-Cost Averaging Means and How It Works.
For more context, read What CAGR Means and When It Misleads.
