CFA Level I Exam · The Behavioral Biases of Individuals
Belief Perseverance Biases in CFA Level I
Updated 7 October 2026 · Fact-checked
Belief perseverance biases are cognitive errors where investors cling to existing beliefs despite new evidence. The five you need are conservatism, confirmation, representativeness, illusion of control and hindsight. To answer an exam question, match the behavior in the stem to the bias, then pick the fix: better information and education.
Understand Cognitive Errors: Belief Perseverance Biases
Behavioral finance splits biases into two groups. Cognitive errors come from faulty reasoning or memory. Emotional biases come from feelings and impulses. Belief perseverance biases are cognitive errors. They show up when people hold on to a belief they already have instead of updating it. The CFA curriculum says cognitive errors are easier to correct than emotional biases, because better information, training and education can help.
There are two kinds of cognitive error. Belief perseverance biases are about clinging to old beliefs. Processing errors are about handling information badly. This topic covers the first group. Anchoring and adjustment, mental accounting, framing and availability belong to the second group.
The five biases:
- Conservatism bias: you update your beliefs too little when new information arrives. You hold the old view and under-react. Example: an analyst only slightly raises her earnings forecast after the company reports earnings far above expectations, staying close to her old estimate.
- Confirmation bias: you look for and favor information that supports your view and ignore or discount information that contradicts it. Example: an investor who owns a stock reads only bullish reports on it.
- Representativeness bias: you classify a new item by how much it resembles a familiar category, and you ignore the underlying odds. Two forms matter: base-rate neglect (ignoring how common the category is) and sample-size neglect (treating a small sample as if it were the whole population). Example: a fund beat its benchmark three years in a row, so you assume it is a skilled manager, even though most funds do not do this.
- Illusion of control bias: you believe you can control or influence outcomes more than you can. Example: an investor trades often because she thinks her actions steer results.
- Hindsight bias: after the fact, you believe an event was predictable and that you knew it all along. You remember the past as more foreseeable than it was.
Conservatism and confirmation are easy to mix up. Conservatism is slow updating: you do see the new data but you weigh it too lightly. Confirmation is selective search and selective weighing: you pick evidence that agrees with you. Look at what the person does with the evidence.
Key formulas to remember
- Conservatism bias
- Old belief kept; new information under-weighted
- Slow updating. Often described as under-reaction to new information.
- Confirmation bias
- Seek supporting evidence; discount contradicting evidence
- Selective search and weighing of evidence that agrees with your view.
- Representativeness bias
- Classify by resemblance; ignore base rates and sample size
- Two forms: base-rate neglect and sample-size neglect.
- Illusion of control bias
- Overestimate ability to influence outcomes
- Shown by believing your own actions steer results, such as frequent trading.
- Hindsight bias
- Past outcomes seen as more predictable than they were
- You remember the past as more foreseeable than it was.
- Mitigation rule
- Cognitive errors: correct with information, education and process
- Emotional biases are harder to fix; you often adapt to them instead.
How to solve Cognitive Errors: Belief Perseverance Biases questions
Use this method for any question that describes an investor's behavior and asks for the bias or the best remedy.
- 1Read the stem and find the specific behavior. Ignore the investor's name, the asset and the numbers.
- 2Decide if the issue is holding on to an existing belief or something else. If it fits belief perseverance, move on.
- 3Ask what the person does with new evidence. Under-reacts to it: conservatism. Seeks only agreeing evidence: confirmation.
- 4Ask whether the person judges by resemblance or small samples: representativeness. Ask whether they think they control results: illusion of control. Ask whether they say the result was obvious afterwards: hindsight.
- 5Eliminate the other two options by naming what bias each would describe.
- 6If the question asks for a remedy, choose information-based fixes: wider research, devil's advocate views, base-rate data, written forecasts, a disciplined process.
Quickest way: Keyword match in 20 seconds
When to use it: Use when the stem describes one behavior and the three options are bias names.
- Scan for the trigger phrase: slow to change forecast (conservatism); only reads supporting research (confirmation); looks like a past winner or small sample (representativeness); thinks trading or actions influence outcome (illusion of control); knew it all along (hindsight).
- Pick the match.
- Check that the other two options do not fit better, then move on.
Common mistakes in Cognitive Errors: Belief Perseverance Biases
Choosing confirmation bias when the investor simply updates too slowly.
Both involve sticking to an old view.
Fix: Confirmation needs selective searching or discounting of evidence. Plain slow adjustment to new data is conservatism.
Calling base-rate neglect a form of conservatism.
Both ignore some information.
Fix: Base-rate neglect is part of representativeness: the investor judges by resemblance and ignores how common the category is.
Treating hindsight bias as a forecasting error made before the event.
The word 'bias' suggests a bad prediction.
Fix: Hindsight bias happens after the outcome. The investor believes the result was predictable.
Labeling illusion of control as overconfidence only.
Both involve excess faith in oneself.
Fix: Illusion of control is about believing you can influence outcomes, often shown by frequent trading.
Classifying these biases as emotional and saying they are hard to fix.
Students blur the cognitive and emotional groups.
Fix: All five here are cognitive errors, largely correctable through better information and education.
Worked examples
Example 1
An investor owns shares of a technology company. She reads only analyst reports with Buy ratings on it and dismisses reports with Sell ratings as poorly researched. Which bias does she most likely show?
A. Conservatism bias
B. Confirmation bias
C. Hindsight bias
Show the solution
- Find the behavior: she picks reports that agree with her holding and dismisses those that do not.
- Test conservatism: that would be slow updating of a belief after new data. She is filtering evidence, not just updating slowly.
- Test hindsight: no past outcome is being described as predictable.
- Selective gathering and discounting of evidence is confirmation bias.
Answer: B. Confirmation bias
Example 2
After a market crash, a portfolio manager says, 'The signs were obvious; I knew it would happen.' His written forecasts before the crash predicted continued growth. Which bias is he most likely showing?
A. Illusion of control bias
B. Hindsight bias
C. Representativeness bias
Show the solution
- Find the behavior: after the event he claims it was obvious, which contradicts his own earlier forecasts.
- This is hindsight bias: the past seems more predictable than it was.
- Check option A: no claim about controlling outcomes appears.
- Check option C: he does not judge by resemblance or small samples.
Answer: B. Hindsight bias
Exam tips
- Match the behavior, not the story. Stems use details like sectors and currencies to distract you.
- Know the pairs that look alike: conservatism vs confirmation, and illusion of control vs overconfidence.
- Remember the group label: belief perseverance biases are cognitive errors and are largely correctable.
- Representativeness has two forms, base-rate neglect and sample-size neglect. Expect a stem that hides this in a fund-manager track-record story.
- There is no penalty for wrong answers, so always answer. If stuck, eliminate the bias that does not fit and choose between the remaining two.
Practice questions from The Behavioral Biases of Individuals
- A client faces a choice between a certain gain of 400 and a 50% chance of gaining 1,000 (otherwise nothing), and also between a certain loss…
- A portfolio manager reads only research articles supporting her bullish view on a energy stock and ignores reports that point to weakening d…
- An adviser notes that a client, after a strong run of gains in her account, begins taking larger and riskier positions, reasoning that she i…
- A portfolio manager attributes all of her past successful stock picks to her own skill and all of her poor results to bad luck. She also tra…
- Which description of the aim of behavioral finance is most accurate?
Cognitive Errors: Belief Perseverance Biases: frequently asked questions
What is the difference between conservatism and confirmation bias?
Conservatism bias means you update your beliefs too slowly when new information arrives. Confirmation bias means you look for and favor information that supports your view and discount information that contradicts it. One is slow adjustment, the other is selective use of evidence.
Are belief perseverance biases cognitive or emotional?
They are cognitive errors. They come from faulty reasoning, not from feelings. Because of that, they are largely correctable through better information, education and a structured process.
What is base-rate neglect in representativeness bias?
Base-rate neglect means you judge an item by how closely it resembles a category and ignore how common that category is. For example, a fund with a strong three-year record looks like a skilled manager, though few funds are. The related error is sample-size neglect.
How do I spot illusion of control in an exam question?
Look for an investor who believes her actions influence outcomes. A typical sign is frequent trading based on confidence that she can steer results. The stem may also say the investor feels in charge of a process that is largely random.