Lesson 1 of 7
How investing judgement fails
Four biases that recur in research, where each one enters, and the written check that catches it.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- Four biases recur in research: overconfidence, anchoring, confirmation and hindsight.
- Each one enters research at a specific moment, not as a general character flaw.
- A written check catches what a private feeling of certainty does not.
Judgement rarely fails as one large mistake. It fails in the same few places, in roughly the same way, for professional and amateur investors alike. Naming the place a bias enters is most of the way to catching it, because a check aimed at the wrong moment in the research process catches nothing.
Four biases and where they enter research
| Bias | Where it enters research | The check that catches it |
|---|---|---|
| Overconfidence | A forecast feels certain because it is yours | Write a range, not a single number |
| Anchoring | A new figure gets read against the price you paid | Value the business with the entry price hidden |
| Confirmation | You stop reading once you find the source that agrees | Look for the one source that would prove you wrong |
| Hindsight | The outcome makes the case that led there feel obvious | Date the case in writing before the outcome is known |
Overconfidence
Overconfidence is mistaking a feeling of certainty for evidence of accuracy. Barber and Odean found that investors who traded the most earned the lowest returns, net of costs, consistent with overconfidence driving activity that analysis alone would not support. The remedy is to write a range around a forecast instead of a single number, and narrow it only when new evidence arrives, not when confidence simply grows.
Anchoring
Anchoring is judging new information against a number that has no claim on the business, most often the price you paid. A stock down from your entry price can feel cheap, and one up from it can feel like proof the thesis was right. Neither feeling is about what the business is worth today; the entry price is a fact about your own trade, not about the company.
Confirmation
Confirmation bias is stopping the search once you find the source that agrees with a view you already hold. A quarter with three ambiguous sentences will supply one to any conclusion; confirmation is choosing which one to read. Catch it before deciding: name the source that would prove you wrong, and read it.
Hindsight
Hindsight bias is the outcome rewriting the story of the decision. Once a company's decline is known, the filing that preceded it reads as an obvious warning, though nothing in it looked obvious on the day it was filed. Kahneman describes this as the mind's tendency to construct the best possible story from what it now knows, losing track of what it did not know at the time.
The four biases feed each other
- 1OverconfidenceA forecast feels certain because it is yours
- 2ConfirmationYou search only for what supports it
- 3HindsightIf it fails, the mistake looks obvious in reverse
Then back to step 1, Overconfidence.
Overconfidence in a forecast invites a search for evidence that supports it, which is confirmation bias doing overconfidence's work. If the forecast then fails, hindsight makes the original mistake look as though it should have been obvious, which discourages the honest look back that would actually help next time. Treating the four as separate checks, run at separate moments, is what keeps one from quietly feeding the next.
Same numbers, read against the entry price
Trading above $50
- Revenue down 3 percent, margins unchanged.
- Reads as a soft quarter, nothing structural.
Trading below $50
- The same revenue and margin figures.
- Reads as the thesis breaking.
Nothing in the filing changed. The anchor was the purchase price, not the business.
Check yourself
An investor keeps citing the one analyst note that agrees with a stock they already own, and skips the two that disagree. Which bias is this?
- Hindsight
- Confirmation
- Anchoring
Show the answer
Answer: B.
Confirmation bias stops the search once a source that agrees is found. The check is looking for the one source that would prove the view wrong.What is the best written check against hindsight bias?
- Trusting your memory of what you expected
- Reviewing the outcome carefully once it is known
- Writing the reasoning down before the outcome is known
Show the answer
Answer: C.
Memory edits itself once an outcome is known. A case written down in advance is the only record hindsight cannot rewrite.A forecast feels certain to you personally, more certain than the evidence behind it supports. Which bias is this?
- Hindsight
- Overconfidence
- Anchoring
Show the answer
Answer: B.
Overconfidence is mistaking a feeling of certainty for evidence of accuracy. A written range, not a single number, is the check.Sources
- Tversky and Kahneman, 'Judgment under Uncertainty: Heuristics and Biases', Science, 1974, on anchoring and adjustment, one of the three heuristics it describes.
- Kahneman, Thinking, Fast and Slow, 2011, on the System 1 thinking behind overconfidence and hindsight.
- Barber and Odean, 'Trading Is Hazardous to Your Wealth', Journal of Finance, 2000, on the cost of overconfident trading.
What this leaves out
- Four biases are named here among many studied in the literature; this is not a complete list.
- Naming a bias does not remove it. A written check does more than awareness alone.
- Nothing here is advice about any security.