How it works
What the model produces
Three numbers per catalyst, built from public regulatory and trial-registry filings, refreshed nightly.
Three numbers per event
- Probability of success — how likely the event is to resolve positively, expressed so that a 70% forecast means roughly seven of ten such events land.
- Two-sided move forecast — what the stock does on a win and on a miss, as separate figures rather than one average. A binary event does not have an average outcome.
- Bias Score — a 0–100 reading of which way the event leans once both of the above are accounted for, shrunk toward neutral when the underlying data is thin.
Where the inputs come from
Trial registrations, FDA approval history, advisory committee notices and sponsor disclosures, all from primary sources, plus cash position and financing activity from filings. Every one of them is a public record.
Every input is listed by name, with its licence terms, on the data sources page. Nothing is scraped from a competitor and nothing is hand-entered.
How often it updates
Every upcoming catalyst is re-scored nightly after the close, and again intraday when a disclosure moves a date. Each score is stored with its timestamp, so a forecast you acted on last month stays on the record as it was published.
What it will not do
Aperture does not produce a buy list, send an alert telling you to go long, or track a portfolio. It measures the event and the price and shows you both. The position, the size and the account are yours.