// STUDY RESULT · 1998-01 TO 2026-09

Inspired by Bernard and Thomas: Post-Earnings Drift (SUE)

NO EDGE

Stocks keep drifting in the direction of a quarterly earnings surprise for about 60 trading days. Bernard and Thomas (1990) trace it to prices acting as if earnings follow a seasonal random walk, ignoring their autocorrelation.

How this run differs from the source: Enters on the 10-Q filing date, often weeks after the earnings release; the paper's drift sits in the days after the announcement.

	THE STRATEGY	$SPXTR BUY & HOLD
CAGR	9.2%	9.3%
Sharpe ratio	0.59	0.66
Max drawdown	-50.1%	-50.9%

// GROWTH OF $1, LOG SCALE

AMBER: THE STRATEGY · GREY: $SPXTR BUY & HOLD

Out-of-sample (from the split): CAGR 9.4% against 14.0% for the benchmark; Sharpe 0.60 in-sample, 0.57 out-of-sample.

After 25 bps of trading costs: 8.1% a year. BELOW RANDOM

// DECADE BY DECADE

ERA	STRATEGY	BENCHMARK
1990s	7.3%	20.9%
2000s	3.7%	-0.8%
2010s	17.0%	14.0%
2020s	7.2%	15.2%

// YEAR BY YEAR, LOSING YEARS IN RED

1998*27.2%27.2%

1999-4.8%21.0%

200039.4%-9.1%

2001-8.9%-11.9%

2002-23.6%-22.1%

200320.1%28.7%

200427.0%10.9%

20055.3%4.9%

20067.7%15.8%

200710.8%5.5%

2008-36.9%-37.0%

200914.3%26.5%

201021.4%15.1%

20117.8%2.1%

201216.5%16.0%

201359.5%32.4%

201414.0%13.7%

20157.1%1.4%

20166.9%12.0%

201723.8%21.8%

20181.8%-4.4%

201919.5%31.5%

202018.7%18.4%

202128.6%28.7%

2022-33.6%-18.1%

202327.8%26.3%

202427.3%25.0%

20250.2%17.9%

2026*-1.2%11.8%

Losing years: 1999 (-4.8%), 2001 (-8.9%), 2002 (-23.6%), 2008 (-36.9%), 2022 (-33.6%), 2026 (-1.2%). Each cell: the strategy, then the benchmark below it. * a part year.

// THE IDEA, AS PUBLISHED

Bernard & Thomas (1989, 1990), JAR/JAE · 1989

SUE = (EPS - EPS four quarters ago - drift) / std dev of past forecast errors
Sort into deciles using the prior quarter's cutoffs; long top decile, short bottom
Hold 60 trading days after the announcement

RUN IT YOURSELF ▸ IN THE LIBRARY

A real computation on the QuantGPT warehouse, walk-forward. A study across the whole market includes only part of the companies that later delisted from 1998 (survivors only before); a study on your own list of tickers is not affected. Gross returns, no costs except where stated. Past performance is not a promise. Findings, not advice. COMPUTED 2026-11-08 19:25 ON THE QUANTGPT WAREHOUSE. COMPANIES THAT LATER DELISTED ARE ONLY PARTLY IN THIS UNIVERSE (A FIX IS IN PROGRESS), SO THE RESULT LEANS TOWARD SURVIVORS. FUNDAMENTALS BY FILING DATE. NOT INVESTMENT ADVICE. PRIVACY