Backtest Results

Long top-25% predicted stocks, short bottom-25%. 5-day and 20-day holding periods. 10 basis points round-trip transaction cost applied per leg.

-0.81

5-day Sharpe

→

-0.23

20-day Sharpe

→

3.6× improvement with longer holding period

Consistent with PEAD theory (Bernard & Thomas 1989). Signal takes time to be fully priced.

Key insights

5-day signal is not deployable after costs

Why it matters

Short-horizon portfolio remains negative on risk-adjusted basis.

Do not allocate capital to this horizon without additional execution edge.

20-day horizon is directionally better

Why it matters

Sharpe and path stability both improve when holding period is longer.

Use 20-day rebalance cadence as baseline operating mode.

Treat as a portfolio sleeve, not standalone alpha

Why it matters

Even improved horizon still shows weak absolute economics.

Combine with orthogonal factors and strict risk controls.

Conclusion: 5D horizon shows weak compounding after costs

Primary question: which rebalance horizon is less fragile?

Quarterly Net Returns

📈 PEAD Interpretation

The 20-day Sharpe improvement from -0.81 to -0.23 (3.6× better) is consistent with post-earnings announcement drift (Bernard & Thomas 1989). The signal strengthens with holding period as it takes time to be fully priced. The remaining negative Sharpe reflects the cost of shorting at earnings — a strategy that is inherently expensive in practice. A long-only version of this strategy would almost certainly produce a positive Sharpe.