Stop Trading the News. Start Trading the Leakage
How AI is revealing the footprints of informed capital in biotech
In the high-stakes world of biotech investing, the Efficient Market Hypothesis (EMH) is a comforting illusion. The theory suggests that all public information is already priced into a stock, making it impossible to consistently beat the market. Yet, every day, traders try to do just that, reacting to FDA announcements and clinical trial results with lightning speed. They’re playing a game they are almost mathematically guaranteed to lose.
This is the uncomfortable truth: the real money isn’t made by reacting to the news. It’s made by anticipating it. While most of the market is busy chasing headlines, a select group of sophisticated investors—the “whales”—are quietly positioning themselves days or even weeks in advance. They aren’t faster; they’re smarter. They’re not trading the news; they’re trading the information leakage that precedes it.
That creates a paradox. The market appears efficient on the surface, but beneath it lies a structural inefficiency, a ghost in the machine that systematically rewards those who can detect its footprint. For the average financial analyst or quant, this presents both a significant risk and a compelling opportunity. The risk is in trading the noise; the opportunity is in finding the signal.
The Signal-to-Noise Catastrophe
The biotech sector is notoriously event-driven. A single piece of news—a successful Phase III trial, an unexpected FDA rejection—can send a stock soaring or crashing. (See Ozempic not too long ago!)
The conventional wisdom is that an edge comes from reacting to these binary events faster than anyone else. This has led to an arms race in low-latency trading, where algorithms fight over milliseconds to process SEC filings.
But what if this entire premise is flawed? This is the question explored by quantitative analyst Glen Carter. (Full disclosure: Glen is a friend of mine, and this piece is not sponsored; I’m just fascinated by the problem he’s solving). Through his research platform, Catalyst Ventures, he argues that most traders are looking in the wrong place.
In a recent deep-dive analysis, his team examined over 8,900 regulatory filings from 169 biotech companies over a three-year period. Their initial findings were sobering: blindly trading on 8-K filings, the very documents traders scramble to parse, yielded a statistically insignificant average return of just +0.21% .
Glen’s research can be summarized as such:
The market is noisy. Most SEC filings are administrative. Blindly trading ‘news’ applies no statistical edge because the signal-to-noise ratio is near zero. For the retail investor, this is the ‘kill zone.’ Trying to trade headlines without differentiating between noise and signal is mathematically guaranteed to bleed capital over time.
On Catalyst Ventures, the key isn’t speed, but context. By segmenting companies into two distinct cohorts—pre-revenue “Clinical-Stage” firms and established “Commercial-Stage” firms—a clear signal emerged from the noise.
Company Stage | Definition | Impact of News (8-K Item 8.01)
------------------------------------------------------------------------
Clinical-Stage | Pre-revenue, R&D-focused | Existential: Average gap-up of +4.94%
Commercial-Stage | Established revenue, stable | Administrative: Negligible move of +0.22%Table 1: The impact of news is highly dependent on a company’s development stage. For clinical-stage firms, news is a matter of survival, whereas for commercial firms, it is often just part of operations. Source: Catalyst Ventures
This distinction is critical, but it’s only half the story. The most powerful insight came from analyzing the days leading up to a major announcement. Carter’s team found that clinical-stage stocks exhibited a stunning +9.4% average run-up in the 10 days before a major positive news event. The “smart money” wasn’t waiting for the 8-K; they were already positioned.
The Academic Backing for “Informed Trading”
This phenomenon of pre-announcement drift is not just a recent anomaly. Academic research has long provided indirect evidence of what many suspect is widespread informed trading.
A seminal study published in the Journal of Investigative Medicine as far back as 2000 found a highly significant difference in the stock price of clinical trial “winners” (+27%) and “losers” (-4%) in the 120 days leading up to the public announcement . The authors concluded that their results “provide indirect evidence that insider trading may be common in the biotechnology industry.”
More recent research reinforces this view. A 2024 event study in the journal PLOS One analyzing over 500,000 news releases found that news related to clinical trials and acquisitions were significant sources of “potential news leakage,” where stock movements preempt the official announcement .
This isn’t necessarily illegal insider trading in the classic sense. The “whales” Carter refers to are often institutional funds with deep scientific due diligence capabilities, expert networks, and proprietary channel checks that give them a significant, entirely legal, information advantage. They are simply connecting the dots better and faster than the rest of the market.
Trading with GenAI
If the alpha is generated in the dark, how can an investor without the deep capacities of a well-staffed institutional fund possibly capture it? This is the problem Catalyst Ventures was built to solve.
As their tagline states, the goal is to “combine state-of-the-art Generative AI with real-time community consensus to decode the complex world of biotechnology investing.” Instead of trying to predict the news, their “Whale Tracker” algorithm is designed to detect the footprint of the accumulation that precedes it.
The platform’s engine uses a unique AI pipeline to sift through immense volumes of unstructured data—from SEC filings and clinical trial databases to breaking news and scientific publications. This is where the Large Language Models (LLMs) come in. They provide the crucial contextual filter, automatically classifying companies and distinguishing between routine administrative filings and potentially market-moving events. This allows the system to execute its three-phase process at scale:
Detect Anomalies: It scans hundreds of biotech micro-caps not for price movement, but for “hidden buying”—significant blocks of volume being absorbed without spiking the price.
Apply AI-Driven Context: The LLM-powered engine isolates signals coming from the high-stakes, Clinical-Stage cohort, filtering out the noise from larger, more stable commercial companies.
Generate a Whale Score: It produces a composite score (0-100) that ranks the conviction of the accumulation, indicating the mathematical probability that the volume is “informed” rather than random.
This hybrid approach anchors a technical signal (volume anomalies) to a fundamental catalyst (a pending clinical trial result). It allows investors to position alongside the informed money during the pre-event run-up, capturing the alpha that most traders, who are waiting for the news to hit the tape, will miss entirely. In other words, if you’re an individual trader or a small investment fund, you get to spy on the giant investors and reap similar rewards as they do.
How to Get Early Access
Catalyst Ventures is currently in beta and looking for sophisticated investors, quants, and analysts to join as test users. This is an opportunity not only to gain insight into a cutting-edge application of AI in trading but also to actively shape the development of a powerful new tool.
The platform provides the data and the signals; users provide the wisdom and feedback that will refine the system. If you are curious about exploring the edge that generative AI can provide in one of the market’s most complex sectors, you can create an account and join the beta program at https://catalyst-ventures.eu.
(Beware that there might still be small bugs! Feedback is thus highly encouraged and can be sent directly to Glen at glen.christopher.carter@gmail.com.)
The Bottom Line: From Noise to Signal
Platforms like Catalyst Ventures represent a crucial shift in investment strategy. They move beyond the futile arms race for speed and instead focus on a more sophisticated challenge: separating signal from noise.
For quantitative analysts, this provides a new dataset to model—the behavior of informed capital. For risk managers, it offers a framework for understanding that the greatest danger isn’t volatility, but trading on meaningless information.
And for the impact-driven investor, it addresses a fundamental issue of market fairness. By making the footprints of the “whales” visible, it helps to level a playing field that has long been tilted in favor of a small circle of insiders.
The biotech market isn’t efficient in the way textbooks describe. It is a complex, noisy ecosystem where information asymmetry is a powerful and persistent force. The winning strategy is not to be the fastest reader of the news, but the most discerning reader of the silence that comes before it.
References
[1] Carter, G. (2026). Quantifying the Edge: Statistical Arbitrage in Clinical-Stage Biotech Events. Catalyst Ventures.
[2] Overgaard, C. B., et al. (2000). Biotechnology stock prices before public announcements: evidence of insider trading? Journal of Investigative Medicine, 48(2), 118-24.
[3] Cho, J., et al. (2024). How does news affect biopharma stock prices?: An event study. PLOS One, 19(1), e0296927.
Reads of the Week
This deep dive by Nancy L. Parenteau, Ph.D. into Revolution Medicines reveals how biotech success is often less about a flawless initial strategy and more about the ability to evolve. Starting as a chemistry-led venture, the company navigated a complex path—divesting non-core platforms, pivoting from failed partnerships, and ultimately committing to a bold and previously “undruggable” target in oncology. For anyone curious about how biotech companies find their strategic identity—not through hype, but disciplined decision-making—Revolution Medicines offers a compelling blueprint.
Corin Wagen argues that biotech is undergoing a horizontalization shift similar to what transformed the software industry—moving from monolithic companies doing everything in-house to ecosystems of specialized vendors. As drug discovery grows more complex and AI-enabled, companies like Plasmidsaurus, Cradle, and Adaptyv are carving out focused, service-based roles that help others innovate faster and more efficiently. This essay highlights how distributed expertise, not vertical empires, may be the future—creating more opportunities, preserving institutional memory, and fostering a healthier, more collaborative innovation landscape.
This thoughtful 2025 biotech reflection on The Brave Brand highlights a crucial yet under-discussed issue: the growing gap between scientific innovation and the systems meant to support it. While biotech continues to attract capital and push boundaries—particularly in radiopharmaceuticals—bottlenecks in clinical trial infrastructure, patient access, and human systems are quietly stalling progress. Breakthroughs alone aren’t enough—trust, coordination, and investment in the “unsexy” parts of the pipeline are just as vital if we want science to truly reach the people who need it.




Grateful for the link, Ari. I appreciate how you’re thinking about where information actually leaks into systems.