Four studies, one pattern: the more optimistic the call, the less you should trust it.
When Goldman Sachs, JPMorgan, or any other major bank publishes a 12-month price target, it reads like a forecast. The research says we should treat it as something closer to a position statement — informative about what the bank is willing to put its name behind, far less informative about where the stock will actually trade.
The Track Record
Bradshaw and Brown (Harvard Business School / Georgia State, 2006) examined roughly 100,000 twelve-month price targets issued between 1997 and 2002. By the end of the twelve months, the stock had reached or exceeded the target in only about a quarter of cases. Even allowing for the target being touched at any point during the year — a much looser bar — it happened less than half the time. A separate study, Asquith, Mikhail and Au (2005), found a similar pattern: targets were achieved at some point within the year in 54.28% of cases.
The Direction Problem
Hitting the exact number is one thing. Getting the direction right is a lower bar, and even there the record is weak. Lee, Miao and co-authors (2024, International Review of Economics & Finance) found that only 54% of targets correctly predicted whether the stock would rise or fall — barely better than a coin flip. The same study documented a systematic upward bias of 9.4% and an average absolute pricing error of 24.8%.
This sample was an emerging market, not US large-caps — worth flagging, since it likely overstates the problem for the most liquid, heavily covered names, and understates it for smaller or less-followed ones.
Why the Bias Runs One Direction
Kerl and Walter, studying German stocks, found something specific: the further a target sits from the current price, the less accurate it tends to be, ex post. The most aggressive, most optimistic calls are precisely the ones the literature says to trust least.
A separate strand of research — Dugar and Nathan; Lin and McNichols; Michaely and Womack — ties the upward bias itself to a structural incentive: analysts at banks with underwriting or advisory relationships to the company they cover have something to lose by publishing an unfavorable number. Losing management access, or future deal flow, is a real cost; being wrong about a price target a year later rarely is.
A Rare Call
Sell ratings remain rare today. As of December 2025, FactSet counted 12,696 analyst ratings across S&P 500 stocks: 57.5% Buy, 37.7% Hold, and just 4.8% Sell. That figure has sat in roughly the 5–6% range for years — a small fraction of all coverage, regardless of where the market itself was heading.
We won't claim that scarcity makes a sell call more accurate — we don't have a study that tests that directly, and we'd rather say so than invent one. What we can say is that when a bank does go negative, it's choosing to issue the rating its own incentive structure pushes against. That alone makes it worth a second look, even without a verified accuracy edge attached to it.
Our Own Rule
None of this means ignore Wall Street. It means reading a target for what it is: one institution's public position, shaped by incentives that don't always point toward accuracy. We treat a price target the same way we'd treat a single data point in any model — useful in context, useless as a conclusion on its own. The temptation, especially when a target is far above the current price, is to let the number do the thinking. That's exactly the case the research says to be most careful with.
A Sentiment Reading, Not a Forecast
A price target tells us what a bank is willing to publish about a company it often has a commercial relationship with. It is a data point about sentiment — not a forecast we should weight as a probability.
