Summary

Verdad’s core claim is that a systematic, quantitative approach works in biotech — the one sector quants have historically written off entirely. Dan Rasmussen’s starting puzzle: biotech is ~25% of the Russell 2000, the least correlated sector in the market, and yet screens like Joel Greenblatt’s “just exclude the entire biotech sector.” Greg Obenshain admits he initially ignored the assignment — “I thought there’s no way you can build a quantitative strategy in biotech” — before the results converted him.
The headline signal is specialist-fund consensus: treat the ~70 funds with >50% of their portfolio in biotech as “a voting machine.” Names owned by multiple specialists, relative to total fund ownership, outperform strongly; companies owned by lots of generalist funds but zero specialists “do terribly.” Counterintuitively, specialists’ unique ideas underperform their consensus ones — “their alpha opinion where they’re the only guy that thinks it is probably wrong” — so copying agreement beats copying conviction.
Verdad’s value metric flips value investing on its head: the anchor of worth is spend — the gap between revenue and cash flow from operations — against market cap. The “dumb insight”: a company that spent $500M on clinical trials is on average worth more than one that spent $10M at the same market cap. Despite its crude construction, “the value metric works better than the specialist metric” and is one of the most powerful return drivers in the model.
On the short side, biotech beta is structurally bad — 60–70% of individual stocks lose money — but concentrated fundamental shorting has burned specialists into abandoning it (running ~105% long, 5% short in XBI). Rasmussen’s answer is quant risk management: “I’d rather be short 70% of the biotech stocks individually than seven of my highest-conviction ones.” Returns come more from slow bleed than blow-ups, and shorts add value by dampening volatility drag — “you can actually lose a little bit of money on your shorts… and be far better off for having shorted.”
Insider buying is predictive — but the useful signal is largely ex-CEO: “the CEO always buys whether or not the stock does well,” while CFO and other C-suite purchases carry signal that persists for months. Sales tell you little since everyone sells issued stock; Verdad mostly counts buyers rather than dollar volume, and boards are only “mildly predictive.”
Classification and momentum run through clinical-trial data: companies are described by aggregating their trials into time-series similarity indices, and momentum works thematically within those peer groups. Andrew Walker’s pushback — a competitor’s approval should hurt you — gets Rasmussen’s Toyota/Ford rebuttal: macro themes such as obesity drugs and mRNA can outweigh isolated competitive events, and “most of the movements in biotech are not event specific.”
The meta-position: Verdad hunts bombed-out sectors, and biotech specialists’ own pessimism (“why would you guys look at biotech?”) was the tell — “like oil in 2015 or 16.” Rasmussen’s frustration now is that his contrarian calls went consensus: “private equity is in a bubble, stay the f out of it” is on the front page, Japan has been among the best-performing markets, and the firm is struggling to refresh its basket — jokingly eyeing private-credit BDCs at big discounts.

Deep dive

1. Biotech is 25% of small cap, uncorrelated, and ineligible for traditional screens — which is exactly why Verdad attacked it

Rasmussen’s setup: as a value investor biotech is “totally ineligible… they’re all money losing. So you’re like, but should I just write this entire sector off, but it’s 25% of the Russell 2000, so like what the hell do I do?” Walker adds that Greenblatt’s magic formula — and “tons of quants” — simply exclude the sector, which is precisely the opportunity.
The portfolio logic: biotech is “the least correlated sector. It’s really weird” — a huge, systematic-free source of uncorrelated return. Rasmussen asked his bond specialist to take on the year-long project on the joke that “biotech stocks are just like bonds, I guess” — there’s a lot of cash in both.
Obenshain’s conversion arc, worth keeping: “You can’t really use the financial statements… at first I kind of ignored it because I thought there’s no way you can build a quantitative strategy in biotech.” The unlock was meeting “biotech on its own terms — only use factors that actually made sense in the realm of biotech.”

2. Specialist consensus is a voting machine — and lone-wolf conviction is a negative signal

The definition is deliberately simple: a specialist is any fund with more than 50% of its portfolio in biotech — currently ~70 funds, unweighted by track record (ARKG counts as “one data point, one vote”). Strategic pharmaceutical holders such as Pfizer can also appear on the specialist list, though strategic-ownership data is “messier” and remains in the research pipeline. Walker summarizes the simplifying assumption: someone gave these funds money, so they presumably know something about the sector.
The construction that matters: not raw specialist count but specialists relative to all funds that own it. The kicker Obenshain calls “really fun”: “companies that are owned by a lot of funds but zero specialists — they do terribly.” It ends up functioning as a quality screen.
Rasmussen’s answer to “why not just invest with the specialists?”: when specialists have unique ideas, they don’t do as well. “The consensus is the signal… you don’t want your biotech manager to have independent opinions. Their alpha opinion where they’re the only guy that thinks it is probably wrong.” Walker’s gloss: they may have found the one sector where a best-ideas fund actually works.

3. The signal isn’t event prediction — it’s better-behaved stocks and a slow-moving 13F

Obenshain’s honest non-answer on mechanism: “I don’t have a direct answer for you” on whether it’s trial success or M&A — though highly-owned names “tend to get acquired,” and acquisition “is the measure of success largely in this industry.” But there aren’t enough events for event studies; the tradeable fact is that specialist-owned names “behave better — higher returns relative to their volatility,” ideal inputs for a constantly rebalanced risk model.
On turnover risk from 13F lag, Obenshain’s rebuttal: a billion-dollar fund with $50M in a small-cap biotech is “completely illiquid… you’re stuck.” Consensus positions can’t move fast — “the chances that 10 different biotech funds all sell simultaneously in a small cap… it’s going to tank the price.”
Walker’s structural objection — specialists can receive better terms through PIPEs and penny warrants, with beneficial-ownership filings sometimes revealing stakes that are not obvious from ordinary ownership data — draws a candid “we’re working on it. I don’t have an answer yet, but we will.” Reader questions feed directly into Verdad’s research pipeline.

4. Shorting biotech: diversify ruthlessly where fundamental managers got annihilated

The premise: “biotech beta is bad” — it’s the sector with the largest share of individual stocks that lose money over time, “maybe 60 or 70%… science projects” with a few lottery tickets. So shorting has to be a component, yet specialist funds have largely abandoned it — running “105% long, 5% short, and the 5% short’s in XBI” — because concentrated, high-conviction shorts got them “annihilated” by promotional pops.
Rasmussen’s fix is pure quant: size by market cap, liquidity, short interest, and borrow cost. “I’d rather be short like 70% of the biotech stocks individually than seven of my highest conviction ones.” The short-interest factor itself works across every sector — stocks with very high short interest and borrow costs “just do terribly” — but the best shorts carry borrow costs that can neutralize expected return, so the model may prefer 50 mildly negative, cheap-to-borrow shorts over 10 expensive high-conviction ones.
Returns on the short book are “more the slow bleed” than sudden drug-failure drops, and the real function is volatility management: “you can actually lose a little bit of money on your shorts over time… and be far better off for having shorted.” Walker’s tail-risk nightmare — SAVA going from $8 to $180 overnight, a 25x move with no time to cover — gets a two-word answer: “Highly diversified.”

5. Insider buying works — but discount the CEO and watch the CFO

The asymmetry: sales are noise (“everybody sort of sells their stock”), non-routine buys are the signal — except “the CEO always buys whether or not the stock does well.” Verdad counts executives ex-CEO; “the CFOs are pretty bearish people, so when they start buying it’s a pretty decent signal.” Boards are only mildly predictive.
Rasmussen’s finding is that insider buys have “power for months after you could observe it” — not a day-trading signal. The literature obsesses over buys before earnings beats, but “that’s not how people in a company think… This is their livelihood. If they think the company’s good, they’re buying it because they think the company’s good.”
Walker notes that biotech insiders face long blackout windows and says he has seen CFOs buying after clinical-data news, including after negative data. The discussion treats the signal as persistent and contextual, not as a phase-specific trading rule.

6. Value redefined: spend is the anchor, and it’s the strongest factor in the model

Obenshain’s definition: spend is “the gap between revenue and cash flow from operations… whether it’s called R&D, whether it’s called SG&A, whether it’s called printer costs, doesn’t matter.” Rasmussen’s framing of the “dumb insight”: assume constant ROI on all biotech spending — a company that spent $500M on trials is presumably worth more than one that spent $10M at the same market cap, even though traditional value metrics rank the bigger loser as worse. Even a failed trial may leave the people behind it trying to repurpose the research.
The punchline Rasmussen volunteers unprompted: “the value metric works better than the specialist metric… one of the most powerful return metrics we have, despite the fact that it’s an incredibly simple construction” — because it measures how value changes over time, which the model can react to.
Walker says his own interest came when pharma was imploding and companies traded below cash; Rasmussen connects that entry point to Verdad’s broader preference for bombed-out sectors: “the more pessimistic people are about something, probably the more interesting of an opportunity.” The specialists they interviewed asked “why would you guys look at biotech?” — “like oil in 2015 or 16.” Walker’s worry that spend tilts toward expensive oncology/Alzheimer’s trials gets “it kind of comes out in the wash,” plus the risk model won’t concentrate the portfolio in any one indication anyway.

7. Classifying companies by trial similarity — and why momentum is thematic, not competitive

The classification problem: biotechs change phases, targets, and lead trials over time, so Verdad builds time-series descriptors by aggregating each company’s clinical trials — then scores similarity to every other company (“I’m like 54 similar to that one, 98 similar to the other one”) to construct a peer index against which value and momentum are measured. Rasmussen calls the phase-3-Alzheimer’s-plus-ten-preclinical-oncology company a “really hard problem” — “this is why quant didn’t want to attack biotech.”
Walker’s momentum pushback: a rival’s bladder-cancer approval should hurt you — at best a duopoly. Rasmussen’s rebuttal: discrete competitive events are the minority; “if Toyota’s going up and you’re Ford, you’re probably going up.” The goal is capturing thematic flows — “as long as people like obesity drugs, I want to own obesity” — and Walker concedes the point with his own example of Pfizer’s Metsera bidding war lifting every obesity name.
Obenshain’s point is that biotech people are “so event-focused. Biotech is events,” but “most of the movements in biotech are not event specific.” The quant edge is targeting underlying drivers rather than predicting readouts.

8. Base rates over predictions — and the frustration of contrarian calls going consensus

Rasmussen’s philosophical close, citing the value-factor literature: news hurts expensive stocks and helps cheap ones because “people think the world is much more predictable than it is.” If the market prices a trial at 90% success, “that’s probably not so smart” — the whole approach is positioning “so that the events end up playing in your favor” rather than handicapping individual readouts.
On the most bombed-out sector today, Rasmussen is stuck: “we’ve said private equity is in a bubble, stay the f out of it” — and now “nobody yells at me like you’re crazy anymore because everybody agrees with me.” Japan, the other call, has been among the best markets. “For a value investor, you start to get worried when you have two or three good years in a row” — hence the half-joke about becoming “huge bulls on private credit BDCs,” which are trading at big discounts.
A final methods note off Walker’s KKR $25M insider-buy question: in the insider analysis, Verdad mostly relies on counts rather than dollar volume — “I don’t want to credit somebody more because they just happened to have more money.” In response to Walker’s ServiceNow example, Rasmussen asks whether anybody else is following the CEO’s lead.

Harvard Prof Reveals Age-Reversing Science to Look & Feel Younger w/ David Sinclair The 2028 Global Intelligence Crisis: Will AI Lead To A Market Crash?