High Phase 3 failure risk

Bearish

A pharma ticker's most urgent Phase 3 trial has a Bayesian failure probability well above the base rate.

PRIMARY SOURCE
Hay 2014
DiMasi 2016
TYPICAL HORIZON
until primary completion (typically 3-12 months)
FACTORS USED
none (universe baseline)

What it means

A biotech or pharma company with a single-asset pipeline where the next Phase 3 readout will dominate price action for the coming year. Our Bayesian engine combined the base failure rate (~42% for Phase 3) with signals like enrollment velocity, endpoint amendments, mechanism of action risk, and cash burn, and arrived at a failure probability materially above the prior.

Why it works

Hay 2014 industry data show ~58% of Phase 3 trials fail. The LLR-based combination of live signals lets us update that base rate with current trial-specific evidence — enrollment anomalies, endpoint amendments mid-trial, historically high failure-rate mechanisms of action. The resulting posterior is directly tradeable because Phase 3 readouts cause ±30-50% price moves and dominate the ticker's next 6-12 months.

Watch out

Binary event. If the trial actually succeeds, the bearish thesis reverses completely and the ticker typically gaps up 30-80% on readout day. This pattern is about probability-weighted expected return, not directional certainty. Position-size accordingly.

Live matches

9 tickers firing right now
TickerCompanySectorChangeScore
BMYBristol-Myers SquibbHealthcare+0.54%43[589]LLYEli Lilly and CompanyHealthcare-1.67%40[586]MRKMerck & Co.Healthcare+1.06%39[585]BSXBoston ScientificHealthcare+6.15%38[584]VRTXVertex PharmaceuticalsHealthcare-0.08%38[584]INCYIncyte CorporationHealthcare-0.13%36[582]NOVO-B.CONovo NordiskHealthcare-0.98%32[578]NVONovo NordiskHealthcare-1.03%32[578]BMRNBioMarin PharmaceuticalHealthcare-4.13%29[575]
Disclaimer. Pattern matches are research signals, not investment advice. Past performance of an academic effect does not guarantee future returns. Forward-return tracking for Framler's own implementation begins 2026-05-16 after the calibration window closes.
More from Framler
Pattern libraryHow patterns compose with FramlerFramler framework paperPer-pattern hit rate (post-16-May)Sector rotation overviewMath invariants live