Anchoring and confirmation bias: how investors fool themselves

Two biases decide a surprising amount of what investors believe about a stock before they have looked at the business. Anchoring is relying too heavily on a first number, such as the price you paid, last year's high or an analyst's target, and adjusting away from it too little. Confirmation bias is seeking out, favoring and remembering information that supports what you already think.

They work as a pair. An anchor tells an investor what a stock "should" be worth; confirmation bias then filters the news so that the anchor is rarely challenged. This article explains both with the original experiments, shows the evidence that anchors move real market prices, works through a real chart, and lays out the habits that counter them. For the wider map of investor biases, see Cognitive biases in investing: what the research actually shows.

Anchoring: the original experiments

In their 1974 paper on judgment heuristics, Amos Tversky and Daniel Kahneman described a simple demonstration. Participants watched a wheel of fortune, numbered 0 to 100 but rigged to stop at either 10 or 65. They were asked whether the percentage of African countries in the United Nations was higher or lower than that number, and then to estimate it. The median estimate was 25 among people who saw 10 and 45 among those who saw 65. The anchor was visibly random, and it still moved the answers by 20 points.

The effect is not confined to students and trivia. In 1987 Gregory Northcraft and Margaret Neale gave real estate agents information about a house that was identical except for its listed price. The agents' appraisals moved with the list price, yet most denied that the listing had influenced them. In a 2006 study by Birte Englich, Thomas Mussweiler and Fritz Strack, experienced German legal professionals rolled loaded dice before recommending a sentence for a hypothetical shoplifting case, and those who rolled higher numbers recommended longer sentences. Expertise does not switch the effect off.

Psychologists explain anchoring in two complementary ways. Insufficient adjustment: people start from the anchor and move away from it, but stop too early. Selective accessibility: considering whether the answer could be near the anchor brings to mind evidence consistent with it, which then shapes the final estimate. The second explanation is where anchoring meets confirmation bias.

The anchors investors use

The purchase price is the most personal anchor, and through loss aversion it drives the disposition effect: selling winners because they are "up" and holding losers until they "get back" to the price paid. The market does not know or care what any one investor paid. See Loss aversion: why losses hurt twice as much as gains please for that mechanism.

Evidence that anchors move market prices

If anchoring only affected individuals, prices might still be set sensibly by the market as a whole. Several studies suggest anchors leave marks on prices too.

The 52-week high. Thomas George and Chuan-Yang Hwang showed in 2004 that stocks trading close to their 52-week high tended to keep outperforming stocks far below theirs, and that nearness to the high explained much of the well-known momentum effect. Their interpretation was anchoring: investors treat the 52-week high as a reference and are reluctant to bid a stock above it even when good news justifies a higher price, so the price adjusts too slowly and catches up later.

Takeover offers. Malcolm Baker, Xin Pan and Jeffrey Wurgler found in 2012 that offer prices in mergers and acquisitions cluster around the target company's 52-week high, a price with no direct link to the target's value, and that the chance a deal succeeds jumps when the offer clears that level. Boards, bidders and shareholders all appear to use the old high as a benchmark of fairness.

Economic forecasts. Sean Campbell and Steven Sharpe reported in 2009 that professional consensus forecasts of US economic releases were anchored toward recent past values, and that markets partly anticipated this bias when the actual numbers came out.

A real chart: PayPal's old high

PayPal's monthly close peaked at $291.48 in June 2021. By the end of 2022 it had lost more than three-quarters of that value, and in October 2026 it stood at $56.55, 81% below the high. An investor anchored on $291 might have called the stock a bargain at $200 in late 2021, at $100 in 2022 and at $70 in 2023. Each of those prices was a long way below the anchor, and each was still well above where the stock was later.

The dashed line shows the trailing 12-month high, the anchor most quote screens display. It resets as old highs drop out of the window, which is one reason a stock can look "cheap" for years: the reference keeps moving down after the price.

Whether PayPal at any of those prices was cheap or expensive depended on its earnings, growth, competition and balance sheet at that time, not on the 2021 price. The chart is not a judgment about the company. It shows how little a past price says about value.

Confirmation bias: the original experiments

In 1960 the psychologist Peter Wason gave participants the number sequence 2, 4, 6 and asked them to discover the rule that generated it by proposing other sequences. Most people formed a hypothesis, such as "numbers increasing by two," and proposed only sequences that fit it (8, 10, 12), receiving a "yes" every time. Few tried sequences that could prove their idea wrong. The real rule was simply "any increasing numbers." People tested to confirm, not to falsify.

In 1979 Charles Lord, Lee Ross and Mark Lepper showed something more troubling. They gave supporters and opponents of capital punishment the same two studies, one supporting and one undermining the claim that the death penalty deters crime. Each side rated the study that agreed with it as more convincing and found flaws in the other, and both sides reported holding their original views more strongly afterward. The same evidence pushed people apart. Raymond Nickerson's 1998 review collected decades of similar findings and called confirmation bias perhaps the best-documented of all reasoning errors.

How it shows up in investing

• Selective search. After buying a stock, investors read more of the bullish coverage and follow the commentators who share their view.
• Asymmetric scrutiny. Good news is accepted at face value; bad news is checked for errors, motives and "noise."
• Selective memory. The calls that worked are remembered; the ones that did not are forgotten or reinterpreted.
• Echo chambers. Online communities organized around a stock reward agreement. A 2013 study of Korean stock message boards by JaeHong Park and coauthors found that investors showed a strong preference for messages confirming their existing beliefs, and that those with stronger confirmation bias were more overconfident, traded more and earned lower returns.

Thesis drift

The most expensive form of confirmation bias in investing is thesis drift. An investor buys for one reason, say rapid revenue growth. Growth slows. Instead of selling, the investor adopts a new reason to hold: the company is now "a value stock," or "a takeover target," or "misunderstood by the market." Each new reason is plausible, and each is chosen because it supports the decision already made. Combined with an anchored purchase price, thesis drift explains how small positions turn into large losses held for years.

Countering both biases

The research on debiasing is less encouraging than the research on biases, but a few techniques have support, and most share a principle: make the disconfirming case explicit before the decision.

1. Consider the opposite. In a 1984 study, Lord, Lepper and Elizabeth Preston found that asking people to imagine how they would judge the evidence if it had pointed the other way reduced biased assimilation more effectively than simply telling them to be fair. For an investor, that means writing the bear case first.
2. Write kill criteria in advance. Before buying, record the thesis and two or three specific developments that would show it is wrong, for example "revenue growth below 10% for two quarters." Review against those criteria, not against the price. A Trading plan and trading journal: what to write down keeps the record honest.
3. Do a zero-based review. Periodically ask of each holding: "If I held cash instead, would I buy this today at this price?" The question removes the purchase-price anchor.
4. Use several anchors deliberately. Value a company with more than one method, such as multiples, discounted cash flows and peer comparisons, so no single number dominates. The spread between the answers is a useful measure of uncertainty.
5. Seek out the best opposing argument. Read the most careful critic of a holding, such as a short seller's report or a skeptical analyst, rather than the weakest.
6. Separate research from position. Some investors write their analysis before deciding whether to buy, so the conclusion cannot shape the evidence.

How Finabase helps

Each symbol page has a Notes tab where signed-in users can keep a written thesis and exit criteria next to the quote (for example PYPL's notes), which makes it harder for a thesis to drift unnoticed. The Valuation tab (PYPL valuation) shows up to ten fiscal years of valuation multiples, a better reference than any past price. The quote header's 52-week range is shown for context, and the News tab and news feed collect coverage from many sources, so it is easy to read both sides of a story.

Frequently asked questions

What is anchoring bias in investing?

Anchoring bias is relying too heavily on a reference number, such as the price you paid, a stock's past high or an analyst's target, when judging what an investment is worth. Investors adjust away from the anchor too little, so an irrelevant number shapes their decision.

What is an example of confirmation bias in stocks?

A typical example is an investor who owns a stock reading mostly bullish analysis, following only commentators who like the company and dismissing negative news as noise, so their confidence grows even as the evidence weakens.

Is the 52-week high a meaningful level?

It is a widely watched reference point, and research by George and Hwang (2004) found that stocks near their 52-week highs tended to keep outperforming, which they attributed to investors anchoring on that level. It is not a measure of value: a stock far below its high is not necessarily cheap.

How can I reduce confirmation bias?

Write the opposing case before deciding, list specific facts that would prove your thesis wrong, read the strongest critics rather than the weakest, and review each holding by asking whether you would buy it today with fresh money.

Summary

Anchoring makes a first number, often a purchase price or a past high, shape judgments about value long after it has stopped being relevant, and research shows the effect reaches market prices and takeover offers. Confirmation bias then filters information so that the anchored view is rarely tested, and when both act together they produce thesis drift: new reasons to hold a position that the original reasons no longer support. PayPal's fall from a $291 monthly close to $57 shows how poor a guide a past price can be. The most reliable defenses make the opposite case explicit before acting: a written bear case, kill criteria and periodic fresh-money reviews.