Catalyst detection · Base rate analysis · News pipelineCorporate events — mergers, acquisitions, spin-offs, proxy contests, management transitions, restatements, and material amendments — create dislocations between fundamental value and observable data signals. Our pipeline ingests corporate event data within hours of publication. Natural language processing classifies each event by type, extracts key terms, and assigns a preliminary probability estimate based on historical completion rates for similar transactions.
Every event type has a historical base rate. Merger completions in specific sectors, activist campaign outcomes, restatement impacts — we maintain a comprehensive database of historical outcomes that serves as the prior distribution for new events. When a new catalyst appears, our systems immediately compute the Bayesian posterior: given this specific language, these specific parties, and this specific market context, what is the updated probability of each possible outcome?
Event-driven situations have natural time horizons. A merger spread has a defined close date. An activist campaign has a proxy deadline. We model each situation as a decaying option: the expected informational value changes as new data arrives and deadlines approach. Signal weights adjust dynamically based on the updated probability distribution, and situations are deprioritised when the expected informational value no longer justifies the analytical resources.