Submarine Catalyst scores every upcoming FDA decision using a 7-factor evidence model — indication-specific approval rates, AdCom vote parsing, SEC filing signals, and convexity analysis. Built by a trader who needed it, not a data vendor who guessed.
Every score is a weighted synthesis of seven independent signal layers — not one metric dressed up as many.
Pulls the live PDUFA calendar daily. Every upcoming FDA decision date — new drugs, label expansions, resubmissions — gets queued for scoring automatically.
Blends indication-specific base rates (Hay 2014, Wong 2019) with review pathway type, AdCom vote parsing, and SEC 8-K filing language signals to produce a calibrated approval probability.
Measures asymmetry, not just approval odds. Proximity to PDUFA, price depression from 52-week high, short interest, historical volatility, market cap tier, and liquidity combine into a single convexity number.
PoA and Convexity are blended into a 0–100 final score using self-calibrating weights that update as actual PDUFA outcomes accumulate in the log. ELITE / STRONG / WATCH / WEAK tiers surface the best setups fast.
Each signal is sourced, weighted, and documented. No black boxes. You can audit every modifier on every ticker.
No freemium limitations. No data delays. No upsells. Full scanner output, updated every scan cycle.
Full scanner · All tickers · All signals
Cancel anytime · Billed monthly via Stripe
Submarine Catalyst started as a private tool — a systematic framework for scoring FDA binary events built by someone who needed to trust instruments over intuition. The same discipline that keeps a submarine crew alive in 600 feet of water transfers directly to trading a PDUFA date: no noise, no emotion, no guessing.
Version 13 of the scanner runs seven independent evidence layers derived from published FDA approval rate studies, SEC filing signals, and convexity modeling. It doesn't pick winners — it identifies where the evidence is strongest and where the setup is most asymmetric.
The performance log is built to be filled in after each event. Every outcome updates the model. That's the part most scanners skip.