The thesis
When a company reports earnings that beat or miss expectations, the price reaction does not happen all at once. Bernard and Thomas (1989) documented that after a positive earnings surprise, the stock continues to drift upward for 60+ days; after a negative surprise, it drifts downward. The drift is most pronounced for stocks with the largest standardized unexpected earnings (SUE), and it has persisted for decades despite being one of the most-studied anomalies in academic finance.
Academic basis
Bernard and Thomas's 1989 Journal of Accounting Research paper remains the canonical reference. Subsequent work (Foster-Olsen-Shevlin 1984; Bhushan 1994; Liang 2003) has shown that the drift is amplified by analyst sluggishness, retail investor inattention, and limits to arbitrage in smaller-cap names. McLean and Pontiff (2016) found that post-publication, the PEAD effect has weakened modestly but not disappeared — consistent with a real risk premium plus a residual mispricing component.
How Alpha Suite implements it
The PEAD scanner pulls earnings calendars and computes standardized unexpected earnings when consensus estimates are available. When SUE data is sparse or unreliable, the model falls back to a gap-and-volume proxy: large overnight gaps on elevated volume function as a market-implied surprise indicator.
- Surprise magnitude — SUE thresholds (positive and negative) trigger long and short candidate flags.
- Volume confirmation — the post-announcement volume must exceed a multiple of the trailing average for the signal to qualify, filtering out small earnings-week noise.
- Drift-window scoring — the model targets the documented 60-day drift window, with position sizing decaying as the window matures.
- Institutional ownership shift — secondary check on whether 13F holders are accumulating or distributing into the surprise.
Each PEAD signal carries a take-profit at the upper barrier, a stop-loss at the volatility-anchored lower barrier, and a time-stop matched to the drift horizon.
When it fires
PEAD signals are heaviest during the four earnings windows each year (the second through fourth weeks of January, April, July, and October). The cleanest fires are mid-cap names with large positive SUE and confirming volume. The model pulls back position size in the days immediately around an upcoming earnings event for any open PEAD position to avoid stacking earnings risk on top of drift exposure.
What it does not catch: One-time blowups where an earnings surprise reflects an irreversible change (fraud, accounting restatement, sudden CEO departure). The model treats those as outliers and adjusts via the volatility-anchored barrier model, but the historical statistical edge does not always hold for them.
See live signals on the dashboard
Recon (free) gives you read-only access to today's ranked signals across all 10 strategies. No card, no commitment — sign up to see the model output for yourself.
Open the Dashboard