Anchoring and adjustment is the cognitive tendency to estimate an unknown quantity by starting from an initial value — the anchor — and then adjusting toward the answer, with the adjustment almost always falling short. Introduced by Amos Tversky and Daniel Kahneman in their 1974 Science paper "Judgment Under Uncertainty: Heuristics and Biases," it explains why a number that has no logical bearing on a problem can still pull a final estimate toward itself. The core tension for markets: prices, forecasts, and decisions get tethered to salient but often irrelevant reference points (a purchase price, last quarter's figure, a 52-week high, a round number), so judgments under-react to new information and drift only slowly toward fair value.

How it works (the mechanism)

The classic demonstrations are striking. Tversky and Kahneman asked one group to estimate 8×7×6×5×4×3×2×1 in five seconds and another to estimate the same product written 1×2×3×4×5×6×7×8. The descending sequence produced a median guess of 2,250; the ascending one only 512 (the true answer is 40,320). The first few terms acted as an anchor. In another experiment, subjects spun a "wheel of fortune," then estimated the percentage of African nations in the UN. Those shown a high spin (65) estimated ~45%; those shown a low spin (10) estimated ~25% — even though the wheel was transparently random.

Two mechanisms are debated, and current consensus treats them as a division of labor rather than rivals:

• Insufficient adjustment (Tversky-Kahneman's original): people start at a self-generated anchor and adjust serially, stopping at the near edge of a plausible range. Epley and Gilovich (2006) showed this applies mainly to anchors people generate themselves.
• Selective accessibility (Strack & Mussweiler, 1997): an externally provided anchor triggers a confirmatory search that makes anchor-consistent information more mentally accessible, biasing the judgment. This explains why even implausible or clearly irrelevant external anchors still bite.

How it shows up in markets

Anchoring is one of the most empirically documented biases in finance, appearing at several levels:

• Forecast stickiness. Federal Reserve researchers Campbell and Sharpe (2009) found professional consensus forecasts of monthly economic releases were systematically biased toward prior months' values, producing predictable forecast errors. Notably, they report that bond yields react only to the unpredictable component of the forecast error — i.e., the market largely anticipates and prices out the anchoring-induced part — so the bias is clearest in the forecasters' behavior, not in an obvious free lunch.
• Analyst earnings estimates. Cen, Hilary, and Wei (2013) documented that analysts anchor EPS forecasts on the industry median: firms with low forecast EPS relative to peers get over-optimistic estimates, and vice versa, with associated return predictability.
• The 52-week high. George and Hwang (2004) found stocks near their 52-week high subsequently outperform, attributing it to anchoring — traders treat the prior high as a ceiling and under-react to good news that should push price past it. This is the most cited markets application.
• Investor reference points. The purchase price, round numbers, and prior highs all serve as anchors. The propensity to trade at round numbers is correlated with worse realized performance (Bhattacharya, Holden, Jacobsen and related round-number work), and anchoring on the purchase price is one driver of the disposition effect (selling winners too early, holding losers too long).
• IPO and deal pricing, target prices. Initial valuation ranges and prior trading levels anchor subsequent price discovery and analyst price targets.

Adoption, debate & evidence

Anchoring is broadly accepted in behavioral economics and is one of the few heuristics with strong replication support. A 50-year meta-analysis by Schley and Weingarten ("50 Years of Anchoring," ~2,600 effect sizes) reported a large overall effect (Hedges' g ≈ 0.83 in the version reviewed here) that the authors report remains large after accounting for extensive publication bias. So the phenomenon is real and durable. (Note: this is a working/SSRN paper and the headline effect size has shifted between draft versions — treat the exact magnitude as approximate, not settled.)

The honest caveats are about moderation and market-level efficacy, not existence:

• Incentives and expertise barely help. Meta-analyses find financial incentives for accuracy have little discernible effect, and debiasing interventions show reduced or null impact. Higher cognitive ability and "consider the opposite" prompts (Mussweiler, Strack & Pfeiffer, 2000) attenuate but do not eliminate it.
• Effect-size inflation. A 2025 replication (Li et al., Economic Inquiry) argues many anchoring estimates come from underpowered designs and overstate magnitude — the bias is real but smaller than headline numbers suggest.
• Anomaly ≠ free money. The 52-week-high effect is a documented anomaly, but like all momentum-family strategies it is exposed to crashes, transaction costs, and crowding; behavioral attribution does not guarantee a persistent net-of-cost edge. Treat "anchoring causes return X" claims more skeptically than the lab demonstrations.

Strengths & limitations

When the concept earns its keep: as a diagnostic lens it reliably flags where judgments are being tethered — your own entry price, a stale forecast, a wheel-of-fortune number masquerading as analysis. It is most predictive when the quantity is uncertain, the anchor is salient, and the judge lacks expertise or time. The selective-accessibility finding (external anchors bias even experts) means professionals are not immune.

Where it fails or misleads: anchoring is descriptive, not prescriptive — knowing it exists does not tell you how much a given price is anchored, and it is easy to "explain" any drift post hoc as anchoring (an unfalsifiable trap). The single most common misuse is treating one's own purchase price as informative: the market does not know or care where you bought, yet that anchor distorts hold/sell decisions (the disposition effect). The second is mistaking a backward-looking anchor (52-week high, round number) for a fundamental level.

Sources

• Tversky & Kahneman (1974), "Judgment Under Uncertainty: Heuristics and Biases," Science — original heuristic and multiplication / wheel-of-fortune experiments.
• Epley & Gilovich (2006), "The Anchoring-and-Adjustment Heuristic: Why the Adjustments Are Insufficient," Psychological Science — self-generated anchors.
• Strack & Mussweiler (1997) and Mussweiler, Strack & Pfeiffer (2000) — selective-accessibility model and "consider the opposite" debiasing.
• George & Hwang (2004), "The 52-Week High and Momentum Investing," Journal of Finance (SSRN 1104491).
• Campbell & Sharpe (2009), "Anchoring Bias in Consensus Forecasts and Its Effect on Market Prices," Federal Reserve FEDS / JFQA.
• Cen, Hilary & Wei (2013), "The Role of Anchoring Bias in the Equity Market," JFQA (SSRN 1572855).
• Schley & Weingarten, "50 Years of Anchoring: A Meta-Analysis and Meta-Study of Anchoring Effects" (SSRN 5114456; later re-titled version SSRN 5893864) — large overall Hedges' g (~0.83 in the version reviewed) surviving publication-bias correction; magnitude shifts across draft versions.
• Li et al. (2025), "Underpowered studies and exaggerated effects: A replication and re-evaluation of the magnitude of anchoring effects," Economic Inquiry.
• Bhattacharya, Holden & Jacobsen, "Penny Wise, Dollar Foolish: Buy-Sell Imbalances On and Around Round Numbers" (round-number anchoring).
• Wikipedia, "Anchoring effect"; Investopedia, "Anchoring" — landscape summaries (cross-checked, not primary).

Dispute flags: the insufficient-adjustment vs. selective-accessibility mechanism debate is genuine (modern view: both apply to different anchor types). The magnitude of anchoring (and the tradability of the 52-week-high anomaly net of costs) is contested — lab existence is solid; market-level dollar edge is weaker than behavioral narratives imply.