Representativeness Heuristic: When Similarity Beats Base Rates

Key Takeaways

• The representativeness heuristic judges probability by how closely something resembles a prototype, rather than by its actual base rate.
• De Bondt and Thaler (1985) found prior extreme losers outperformed prior extreme winners over three to five years, consistent with representativeness-driven overreaction.
• Using the most compelling analogue company as the base case substitutes one vivid example for the full distribution of outcomes in that cohort.
• Base-rate neglect inflates the implied probability of hyperscaler outcomes for high-growth companies well above what historical distributions support.

What It Is

Daniel Kahneman and Amos Tversky introduced the heuristic in their 1972 paper "Subjective Probability: A Judgment of Representativeness" and developed it further in the landmark 1974 Science article "Judgment Under Uncertainty: Heuristics and Biases." Under representativeness, people estimate the probability that item A belongs to class B by how closely A matches their mental picture of a typical B, rather than by the statistical base rate of B.

The heuristic is not wrong in every case. Similarity and probability often line up. The problem is that when the two diverge, representativeness pushes decisions in the direction of similarity, and investors pay the price.

The Intuition

Two numbers matter when you judge whether a stock fits a story. The first is how well the story describes this specific stock. The second is how often that story actually comes true for stocks of this kind. Representativeness makes the first number dominate.

Tversky and Kahneman's famous "Linda problem," expanded in their 1983 paper on the conjunction fallacy, illustrates the pattern outside markets. Told that Linda is a philosophy graduate who is outspoken about social justice, most people judged "Linda is a bank teller who is active in the feminist movement" as more likely than "Linda is a bank teller." The conjunction cannot be more likely than one of its parts, but the added detail makes the compound story more representative and so feels more probable.

How It Works

Representativeness produces several systematic errors in financial judgment.

Base-rate neglect. Investors discount how common a pattern is in the full population. A new software company that "looks like" a future hyperscaler gets priced as if the hyperscaler outcome were 20 percent likely, when the base rate in similar cohorts is closer to 1 percent.

Insensitivity to sample size. Three years of strong returns for a fund feel like evidence of skill. Three years is a very small sample for separating skill from noise, but the run looks like the picture of a good fund, so representativeness dominates.

Misconceptions of chance. The gambler's fallacy and the hot-hand fallacy are both children of representativeness. Short sequences are expected to resemble their long-run distribution, and any deviation feels like a signal.

Regression neglect. Extreme past results tend to move toward the mean. Representativeness projects the extreme forward because the extreme itself is what makes the case feel compelling.

In a 1985 paper "Does the Stock Market Overreact?" Werner De Bondt and Richard Thaler documented a pattern consistent with representativeness at the aggregate level. Prior extreme losers outperformed prior extreme winners over the following three to five years in their sample, a result they attributed to investors extrapolating recent performance too far into the future.

Worked Example

Suppose a young company reports 80 percent revenue growth for two quarters, opens international offices, and lands a major customer. Its management deck looks similar to the early-stage decks of three companies that later became ten-baggers.

The representativeness heuristic prices in a significant probability of a similar path. The stock trades at 45 times forward revenue. An investor benchmarks against the three success stories and concludes the multiple is justified.

The missing number is the base rate. Across the full population of companies that hit 80 percent growth for two quarters, landed a marquee customer, and expanded internationally, how many went on to match the three success stories? Studies of high-growth cohorts typically find that a small share of companies carry the entire cohort's long-term returns. The base rate for the path being priced in is a small number, not a large one.

If you allocate on the vivid match and ignore the base rate, you are using representativeness. If you size the position for a distribution of outcomes where the hyperscaler result is one possibility among many, you are using probability.

Common Mistakes

1. Pricing on the best analogue instead of the full population. The one company this one reminds you of is a sample of one. Average outcomes across the full cohort are the honest reference.

2. Equating a vivid narrative with high probability. The more detailed and coherent a story, the more representative it feels, and the less you check the base rate. Detail can lower true probability while raising perceived probability.

3. Reading short streaks as process. Two good quarters, three strong years, or four winning trades are small samples. Representativeness treats them as the true picture of the underlying distribution.

4. Ignoring regression to the mean after extreme results. A top-decile year is less often repeated than the headline implies. Pricing for persistence is the heuristic at work.

5. Stereotyping by sector or style. "All SaaS companies scale at 60 percent gross retention" or "all value names deserve a low multiple" are representativeness rules in disguise. Each company has its own distribution.

Frequently Asked Questions

What is the representativeness heuristic in simple terms? The representativeness heuristic is judging probability by how closely something resembles a familiar prototype, rather than by its actual base rate. A company that "looks like" a past ten-bagger gets priced as if it were one, regardless of how often companies with that profile actually deliver that outcome.

How does the representativeness heuristic affect investment decisions? It inflates the implied probability of compelling outcomes while deflating the weight given to base rates. De Bondt and Thaler's 1985 research found prior extreme losers outperformed prior extreme winners over three to five years, consistent with investors extrapolating recent extreme performance too far forward, exactly what representativeness predicts.

What is a real-world example of the representativeness heuristic? A software company reports 80 percent revenue growth for two quarters, opens international offices, and signs a marquee client. Three successful hyperscalers had similar early-stage profiles. The heuristic prices in a high probability of a similar trajectory. The base rate, how often companies with this profile actually become hyperscalers, is a small fraction of that implied probability.

How can investors guard against the representativeness heuristic? Always pair the pattern match with the base rate. Before using a historical analogue as your reference case, look up the full population of companies that fit the same profile and how outcomes distributed across them. If the base rate for the outcome you are pricing is 1 or 2 percent, the implied probability in the valuation should reflect that distribution, not a vivid match to one successful predecessor.

How is the representativeness heuristic different from legitimate pattern recognition? Legitimate pattern recognition uses base rates: it identifies a pattern that, across the full relevant population, has shown predictive value with a measured hit rate. The representativeness heuristic skips the base-rate step and judges from a single vivid match. The diagnostic: if your estimate would change after learning the base rate across all similar cases, representativeness was driving the estimate before you checked.

Sources

1. Kahneman, D. & Tversky, A. (1972). "Subjective Probability: A Judgment of Representativeness." Cognitive Psychology, 3(3), 430-454. https://www.sciencedirect.com/science/article/abs/pii/0010028572900163
2. Tversky, A. & Kahneman, D. (1974). "Judgment Under Uncertainty: Heuristics and Biases." Science, 185(4157), 1124-1131. https://www.science.org/doi/10.1126/science.185.4157.1124
3. Tversky, A. & Kahneman, D. (1983). "Extensional Versus Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment." Psychological Review, 90(4), 293-315. https://psycnet.apa.org/record/1984-03110-001
4. CFA Institute. "The Behavioral Biases of Individuals." Refresher Readings. https://www.cfainstitute.org/insights/professional-learning/refresher-readings/2026/the-behavioral-biases-of-individuals

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.