Our Research
Clinical trials, estimated before the readout and scored after it
Clinical Evidence Research asks whether a registered trial will meet its primary endpoint. Before the result is public, we record a probability. After the result is public, we score it. Across many trials, the score shows whether a structured reading of the evidence predicts outcomes better than the base rate.
Phase 2 and Phase 3 trials on public registries with a defined primary endpoint and an expected readout date. For each trial: mechanism, design, endpoint, comparator, sample size, and prior results. We read the public record. We do not run trials, recruit patients, or give medical advice.
Each estimate is a probability with a timestamp, recorded before the readout. Outcomes are resolved by rules fixed in advance. Estimates are scored with the Brier score against two benchmarks: the historical base rate for comparable trials and an uninformative forecast of 50 percent. Results below the benchmarks are reported as such.
The same probabilities enter the other two areas. Equity Research uses them to value healthcare companies. Quantitative Research studies trial readouts as market events. One estimate, recorded once, scored once, used three times.
The CE Protocol and the CE Report. The protocol fixes question, sample, resolution rules, and scoring before data collection. The report publishes estimates, outcomes, and scores. The founding projects are CE-1 and CE-2.
Publications
How often registered Phase 3 trials meet their primary endpoint, by therapeutic area and trial design. The reference point for every estimate in CE-2.
Probabilities for upcoming trial readouts, recorded before the result is public and scored against the CE-1 base rate.
