Graveyard / Equity events
Biotech run-up before FDA decisions
Do small biotech stocks rise in the weeks before an FDA decision date?
The idea
Small biotech stocks often rally into an FDA decision date as traders position for good news. Buy about 30 trading days before the date and sell two days before the decision, so the binary outcome is never held.
How we tested it
| Data | Company filings (8-K) that mention a PDUFA date, giving 899 events after an extension rule. Market cap between $100M and $2bn leaves 222 events; 71 were kept as primary trades. Long the stock from 30 trading days before the date to 2 days before, short the XBI biotech ETF. |
|---|---|
| Period | 2012 to 2026. Holdout 2021 to 2026; the earlier years are the development sample. |
| Costs | 60 bp for the stock and 5 bp for XBI, taken off every trade. |
| Holdout | Dev and holdout were fixed in advance. Four variants (T-60, T-30, T-15, unhedged) were corrected for multiple testing, alpha 0.0167. |
| Control | A random twin: the same ticker with 5 random windows in the same year. |
Result
| Holdout events (primary) | 41 |
|---|---|
| Holdout mean net of XBI | −0.10% per event (t −0.03) |
| Holdout median / win rate | −2.30% / 41% |
| Versus random twin, p value | 0.91 |
| Dev sample (30 events) | +3.71% (t 0.64, twin p 0.92) |
| Best twin p across variants | 0.45 (T-60, holdout) |
| Holdout, T-30 unhedged | +2.41% (t 0.70, twin p 0.47) |
| Events with no usable price | about 37% |
Why it failed
The run-up is not there. The development result of +3.7% is noise, made of two halves with opposite signs (−3.5% and +10.9%). About 37% of events could not be priced, mostly delisted names that failed, so the sample leans toward survivors, which would flatter the strategy, and it still showed nothing. The test can only rule out an edge of about +3% per event or more.
What would make it worth retesting
Only with a price source that includes delisted companies and a clean FDA-action calendar.
Past results of a backtest say nothing certain about the future.