RxOdds

Clinical trial intelligence powered by machine learning.

FDA approval probability predictions for 475,000+ drug programs from 35+ data sources.

Built for biotech investors, analysts, and portfolio managers.

Featured Insight

Oncology programs with Breakthrough Therapy designation show 2.4x higher approval rates.

475K+

Drug Programs Scored

104

Predictive Features

96.8%

Prediction Accuracy (AUC)

35+

Data Sources, Updated Daily

How It Works

1

Ingest

We pull data daily from 35+ public sources — FDA filings, ClinicalTrials.gov, SEC EDGAR, patent databases, medical journals, and more.

2

Score

Two XGBoost models score every active drug program for FDA approval probability using 100+ features, with temporal validation to prevent leakage.

3

Analyze

You screen, filter, and rank programs by POS, pipeline NPV, catalyst proximity, or short-alpha signals to find investment-relevant opportunities.

What You Get

Net Alpha Screen

Risk-adjusted pipeline NPV minus patent-cliff erosion, ranked against market cap. The flagship view — pipeline value vs loss of exclusivity, net.

Per-Drug POS Scores

Every drug program scored individually with probability of FDA approval, plus a SHAP-style breakdown of the features driving each prediction.

Catalyst Intelligence

Upcoming Phase 2/3 readouts, PDUFA dates, and AdCom meetings, ranked by proximity-weighted NPV so near-term events get the attention they deserve.

Drug Screener

Filter 475K+ programs by POS, phase, FDA designations, therapeutic area, and clinical trial characteristics. Inline tooltips on every filter.

AdCom Intelligence

FDA advisory committee member profiling. Voting patterns, industry payments, academic credentials, and predicted panel outcomes.

Options & Short Alpha

ATM implied volatility, IV rank, put/call ratios, and a composite short-alpha score combining pipeline weakness, short interest, and cash runway.

Competitive & Regulatory

Drug-vs-drug comparisons across indications, FDA action tracking, and cross-agency approvals (FDA, EMA, PMDA).

Score Simulator

Toggle features (designations, trial design, endpoints) and see POS change in real time. Shareable via URL — useful for investor memos.

Built on Public Data

ClinicalTrials.govFDA openFDASEC EDGARPubMedOpen TargetsEMAPMDA JapanFinnhubFINRACMS Open PaymentsNIH RePORTERFEC

Two XGBoost models (early-stage + late-stage), 104 features, temporal validation. Prediction accuracy 0.968 AUC without label-leaky features — materially above the published industry benchmark.