Post-earnings drift (equities)
PEAD driver decomposition
ResolvedIn depth
Post-earnings-announcement drift — the tendency of a stock to keep moving in the direction of an earnings surprise for weeks afterward — is the core edge of the equities sleeve. A small live sample raised a hard question: was the edge real and broad, or an illusion created by a few outlier trades? We decomposed roughly five years of earnings events (n≈949) to settle it. With a single outlier removed, the live signal looked like noise; but across the full backfilled population, even after removing the largest contributors, the effect held strongly (t-statistic above 4) and showed up in every market regime tested. The conclusion: the drift is a broad, durable, population-wide phenomenon — not a handful of lucky names — though it behaves as a directional (sign) effect rather than a clean magnitude relationship. We held ourselves to a pre-registered four-part 'fully robust' bar and missed on one part, so the result is recorded honestly as favorable-but-inconclusive. The live book keeps accumulating trades, and we are open that it is still young and concentrated. Going live with real capital is a discretionary decision rather than a fixed trade-count gate; the honest significance check we hold ourselves to is the program-level Deflated Sharpe Ratio, which — after haircutting for every strategy we have ever tried — still reads the live track record as not yet statistically proven.
The question
Is the post-earnings-announcement-drift edge durable and broad-based across the population, or driven by a handful of outlier names? The core go-live question for the equities sleeve.
How it was tested
The hypothesis, its exact trigger, the outcome it predicts, and the pass/fail bar were written down and cryptographically hash-locked before any data was tested, then held through a mandatory cooling-off period. The backfill ran exactly once against those frozen parameters — no re-tuning, no curve-fitting — using adversarial statistics (stationary-bootstrap confidence intervals, multiple-testing correction, purged cross-validation with an embargo, and walk-forward out-of-sample splits). The specific trigger thresholds are proprietary and omitted here.
The outcome
INCONCLUSIVE→leans-DURABLE/BROAD (2026-06-03, N=949): ex-top-1% t=3.85[lo1.89], ex-Top-3 t=4.53, top-1% share 30%<50%, every-regime+; NOT outlier-driven. Misses ROBUST_BROAD only on quintile-IC 2/5 (PEAD=sign-effect not magnitude-monotonic; IC14d 0.026 CI straddles 0). SQUEEZE untestable (FMP SI 0% coverage). No sizing change; live sample accumulating (n≈44) — going live is a discretionary operator decision (the n=60 gate was retired 2026-07-01); the program-level Deflated Sharpe is the honest significance check.
The backfill ran and concluded — the hypothesis did not fail, but the result is informational rather than a clean strategy-grade pass. See the outcome below.
Pre-registration record
- Registered
- 2026-05-28
- Pre-registered
- yes
- SHA-256 hash
- #eab039fd…
The hash and timestamp are a contemporaneous, immutable record that the hypothesis and its success criteria were fixed before testing — not chosen with hindsight.