BioTrials — Pro: Clinical Trials → Intelligence | 558K Trials — 1,515 Tickers, 25-Year Archive (1999-2025)

BioTrials Pro — Clinical Trial Events + Labels + Returns (1999–2025) What makes Pro different Pro adds the layers buyers actually need for modeling and backtesting: - event-level trial timelines, - sponsor-to-ticker normalization, - outcome and therapeutic labeling, - FDA linkages, - and an “ultimate” ML-ready table that merges signals into one dataset. What you get (deliverables) Core trial spine - trials_master.parquet: 558,973 trials - sponsor_entities.parquet: 48,112 sponsor entities (ticker mapping) Event & signal layers (documented in pack materials) - trial_events.parquet: 1,271,750 event rows (registrations, results posted, etc.) - prices_daily.parquet: 10,190,565 rows (ticker × date) - biotrials_labels.parquet: 168,131 events with forward returns - catalyst_scores.parquet: trial-level composite scoring - biotrials_pro_labels.parquet: 168,131 rows, 38-column “ultimate” ML-ready table Enrichment tables (examples documented) - biotrials_fda_links.parquet: trial → FDA linkages - sec_insider_signals.parquet: insider activity signals linked to tickers - trial_outcomes.parquet, trial_therapeutic_areas.parquet, sponsor_pipeline_metrics.parquet Coverage & grain - Coverage window: 1999–2025 - Grains: - trial spine: one row per nct_id - event layer: one row per (nct_id, event_type, event_date) - ML labels: one row per event with forward-return labels QA details (how to validate quickly) Clinical trial datasets are usually validated by key uniqueness (nct_id), date coverage, and