November 19, 2026
Fund

How do biotech hedge funds make money?


The notion of biotech hedge funds always felt a little strange to me. Yes, hedge funds are filled with clever people with clever ideas, like trading brick-and-mortar stocks based on satellite imagery showing how many cars are in their parking lots, but how are you going to trade on clinical trial results? Start up a lab? Whether drugs work in humans feels as unpredictable as anything else in biology. Maybe you can depend on the base rate of how often drugs for a particular disease area are approved, but everyone knows that already. Surely there are no dollar bills on the ground here waiting to be picked up.

And yet, hundreds of billions of dollars are managed by hedge funds and investment firms that specialize in biotech. The largest are Baker Brothers Advisors, Deerfield Management, and OrbiMed, each managing over $15 billion. Below them sit dozens more. How can these funds, some of which have been around for decades, possibly know more than the people doing the science?

Well, a surprisingly large number of clinical trial results are actually decently predictable by anyone willing to simply look. Some very expensive trials are run for seemingly no reason at all, all of which have failed incredibly badly in the past and have little reason to succeed in the future. LLMs may have replaced a decent chunk of the work here, but, in the past, hiring a small army of graduate students to compile information was table stakes for any self-respecting hedge fund. One of the most famous of the pack – RA Capital – has an entire division called ‘TechAtlas’ dedicated to the effort. 

But we needn’t rely on LLMs or human ingenuity for this! Insilico Medicine has published a paper on inClinico, a platform that predicts Phase II clinical trial outcomes using embarrassingly simple inputs: ‘target choice’ features (whether this is a good drug for this disease, based on omics data and a biomedical knowledge graph) and ‘trial design’ features (like enrollment size, blinding, number of endpoints, patient age, location, and funder type). Their predictions, which were date-stamped and published on preprint servers before the trials read out, hit 79 percent accuracy across 14 resolved trials.

But then we arrive at the same problem as before: if it’s as easy as training a model or a human on what is essentially public information, how does anybody make money here?

Sometimes the answer is private information. The biotech investing industry spends enormous sums on paid consultations with expert networks. The business model is straightforward. An analyst at a biotech hedge fund wants to understand whether a particular clinical trial is likely to succeed. They pay an expert network $500 to $1,000 or more per hour to connect them with an individual who treats the disease in question, or who has run similar trials, or sits on advisory boards, or is otherwise an ‘expert’. The analyst asks questions, and the expert answers.

The stated purpose is to gain non-obvious insights. All of this is perfectly above board. There is no legal apparatus that prevents you from staring deeply into a clinician’s glassy eyes and asking them whether the trial’s endpoint is well chosen. There is nothing stopping you from taking their moment of hesitation as a sign of something worth trading on. It is a friendly conversation, and the SEC cannot prevent you from making friends.

This conversation also happens to be the same context in which the single largest insider trading transaction in American history occurred.

The main character is Mathew Martoma, who was a portfolio manager at SAC Capital Advisors, a hedge fund dealing in many areas, including biotech. In 2006, through one of the aforementioned expert networks – Gerson Lehrman Group – Martoma connected with Dr Sidney Gilman, a neurology professor at the University of Michigan who chaired the safety monitoring committee for bapineuzumab, an Alzheimer’s drug in clinical trials. Over two years, Martoma arranged dozens of paid consultations with Gilman. Based on Gilman’s optimistic reports about the trial’s progress, SAC built a position worth roughly $700 million in bapineuzumab’s two sponsoring companies. Then, in July 2008, Gilman learned the drug was failing. He told Martoma. Over the following days, SAC liquidated the entire $700 million position and went short. When the results were publicly announced later that month, SAC had netted $275 million in combined profits and losses avoided by selling the long holdings early.

Obviously, Martoma was convicted in 2014 of securities fraud. SAC Capital paid $1.8 billion in fines and pleaded guilty to fraud charges, and rebranded as Point72 Asset Management. Yes! That Point72!

How did they catch this guy anyway? The government, via an entity called FINRA, runs electronic surveillance on all shares traded on the NYSE, Nasdaq, and over the counter, including historical trading patterns. And they noticed something rather archetypical: someone held a massive long position, liquidated the entire thing, and then reversed to a short position all in the span of a few days, right before catastrophic news hit. FINRA flagged this to the SEC, which in turn alerted the FBI, which in turn pressured Gilman, who was 81, facing the end of his career, and potential prosecution, into cooperating in exchange for no jail time.

Now, an interesting question to ask is: could Martoma have gotten away with it? Probably not, given how egregious both his offenses were. But if you talk to people involved who invest in biotechnology, it seems that the thing that led to the Martoma prosecution never stopped in any meaningful sense. It went on before the case, and it has continued since. Expert consultations are more likely to be recorded now and people are more scared to say something incriminating; the SEC is not magic, but it has ruined enough lives to function as a deterrent at the explicit end of the spectrum.

But the game of increasingly subtle information extraction continues, and it is a game hard to enter for those outside of a hedge fund. You can only imagine how much this continues in entirely private, unrecorded conversations, between two friends who went to graduate school together, one who works as a scientist, and another who works in finance.

This all said: it would be unfair of me to malign the biotech hedge fund industry at large because of stories like this, which do genuinely represent edge cases. We can quibble about the relative utility of, say, nanosecond-faster trading, but it feels obviously correct that accurate pricing of biotech companies is a net good for the world. Good drug developers should get money, bad ones should not get money.

But for the most accurate players in this game, it also feels obvious that two things are true. One, that they are enormously prescient at seeing where the future may go in ways that are often opaque to even their peers – consider Baker Brothers’ twenty-year investment in Seagen, one of the first antibody-drug-conjugate companies, an investment which ended up converting $2 billion into $10 billion. And two, at least some of them are the beneficiaries of information that nobody else could ever reasonably have.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *