What is 'buy the rumor, sell the news'?

"Buy the rumor, sell the news" describes a price that climbs while an event is still ahead of it and then falls once the event arrives, because the people who bought the anticipation sell into the announcement. In biotech it is the phrase reached for whenever a stock drops on a trial win or an approval — and used that way it explains almost nothing.

A biotech stock falls on good news when the news is worse than the news the price had already assumed, and that happens for five separate reasons. The result was real but smaller than the figure investors were holding. The primary endpoint succeeded while a safety or secondary number did not. The approval arrived with a narrower label than the one being valued. The company sold new shares into the rally. Or the information had already arrived in pieces, and the announcement only made official what was known.

Only the last of those five is "sell the news" in any strict sense. The phrase gets applied to all of them, which is why hearing it explains nothing. The five are not variations on one idea. They leave different evidence, they say different things about the company, and four of them can be checked against a document rather than guessed at.

First, check whether the news was actually good

Some of these moves are not the puzzle they appear to be. Karyopharm's phase 3 endometrial cancer release on 30 July 2026 stated that the trial did not meet its primary endpoint of progression-free survival. It also reported median progression-free survival of 12.75 months on selinexor against 7.43 months on placebo. Both statements are accurate. A reader who anchored on the two medians, saw a large gap, and skipped the first sentence would spend the rest of the day looking for an explanation of an irrational sell-off in a stock that had simply failed its trial.

So begin by locating the sentence that says whether the endpoint was met, before reading any number. Our guide to what topline data actually contains works through that step in detail. If the endpoint was met and the stock still fell, the four mechanisms below are where the answer is.

Was the result smaller than the number people were holding?

A trial's pass mark is written into the protocol before the first patient enrolls. The market's pass mark is whatever the company and its analysts said out loud in the years beforehand. Nothing obliges the two to agree, and when they don't, a trial can succeed and disappoint in the same sentence.

Novo Nordisk's REDEFINE 1 trial of CagriSema is the clearest recent case. On 20 December 2024 the company reported that the trial had achieved its primary endpoint with statistically significant and superior weight loss against placebo in 3,417 adults over 68 weeks. The same release gave two different headline figures for how much weight, depending on the question being asked: 22.7% if everyone had adhered to treatment, and 20.4% counting the trial as it was actually lived. Against that, Novo's own chief scientific officer had projected at least 25%, and the shares fell about 20%, taking roughly $72bn of market value with them.

The endpoint test was passed. The comparison the market was running, against a number nobody had written into a protocol, was not. Shareholders later sued over the trial's design and disclosures, and in July 2026 a federal judge allowed part of that case to proceed while dismissing the claims about the weight-loss projection itself, describing those forecasts as aspirational. Management guidance sets the market's bar without ever becoming a commitment.

Did the primary endpoint succeed while something else failed?

Viking Therapeutics announced topline results from its phase 2 oral obesity trial on 19 August 2025 under the headline "positive top-line results," with the drug "shown to be safe and well-tolerated" and 99% of gastrointestinal adverse events mild or moderate. The efficacy was real: up to 12.2% mean weight loss after 13 weeks against 1.3% on placebo. Shares fell as much as 44% that morning.

The reason sat lower in the same release. Twenty percent of participants on the drug stopped treatment because of an adverse event, against 13% on placebo; nausea ran 58% against 48%, and vomiting 26% against 10%. Nothing was concealed. "Well-tolerated" is an adjective; the discontinuation rate is a measurement, and for a drug people are meant to take for years, how many of them keep taking it is closer to the commercial question than how much weight the ones who stayed lost.

That pattern, primary endpoint met and a secondary or safety number carrying the move, is why a release can be accurate in every line and still take the stock down.

Was the approval narrower than the one being priced?

An approval is not the binary event the headline makes it look like. What the FDA approves is a label, and the label names who can be prescribed the drug.

Sarepta's gene therapy Elevidys received accelerated approval on 22 June 2023 for Duchenne muscular dystrophy, restricted to ambulatory patients aged 4 through 5, where the company had sought all ambulatory patients. The shares fell about 11%, with analysts pointing at how hard the path to widening the label looked. The drug was approved. The commercial population it was approved into was a fraction of the one the price had been carrying.

Did the company sell shares into the news?

Good data raises the price at which a company can issue stock. A company that needs money generally raises it at that price, which is a reasonable thing for it to do and a dilutive one for people already holding.

Aptinyx reported statistically significant phase 2 results in post-traumatic stress disorder on 19 October 2020 and the stock rose 55%. The following evening the company announced an offering of 12 million shares, roughly a quarter of the shares outstanding, and the stock gave back part of the move the next day. The data had not changed overnight. The number of shares the value would be divided across had. The sequence, the shelf that makes it possible, and where each step lands on EDGAR are covered in when a biotech raises cash after positive data.

Had the news already arrived in pieces?

This is the one that "sell the news" actually describes, and the hardest to confirm.

When Vertex and CRISPR Therapeutics received US approval for Casgevy, the first CRISPR-based medicine, on 8 December 2023, the UK regulator had already authorised the same therapy for the same two diseases on 16 November 2023. By the time the FDA decision was announced, most of the information in it had been public for weeks.

The mechanism in general form: information arrives gradually, the price moves as it arrives, and the formal announcement is the least surprising part of it. Investors who bought specifically to hold through the event have no reason to hold afterwards, whatever the outcome. This one cannot be verified from outside, because no document records what was priced in. That is also why it is such a convenient explanation for moves nobody has checked.

How to tell which one you are looking at

Each of the first four leaves a trace. Work them in order:

• Read the endpoint sentence before any number. If the trial missed, stop; there is nothing further to explain.
• Compare the effect size to what management said publicly beforehand — investor-day decks, conference presentations, earnings-call transcripts. The bar the market used was set somewhere, and usually in public.
• Read the adverse-event and discontinuation figures, not the tolerability adjective. A company describes its own safety profile; the rates are the part that is checkable.
• For an approval, read the indication sentence in the label and compare it to the indication the company said it had applied for.
• Check for a securities filing in the following day or two. An offering announcement arrives as an 8-K and a prospectus supplement, both public.
• What is left after those is the run-up. It is the residual explanation, not the first one.

For the underlying skill this all rests on — trial design, effect size, and how to read a safety table — our guide to how to read a phase 3 clinical trial readout covers the ground in more depth.

What this does not tell you

Identifying the mechanism explains the move. It does not tell you whether the price that followed is right, and the two are easy to conflate on a day when a position has just moved against you. A narrow label can widen. A discontinuation rate in a 13-week trial may not be the rate in a longer one. Dilution is a fact about the share count, not a verdict on the drug.

What the exercise buys you is knowing which question you are actually asking: one about the data, the label, the balance sheet, or about who owned the stock last week. Those have different answers, and different next dates.

Nothing here is investment advice.