Quant-Driven Earnings Surprise Strategy: May 2026 Stock Selection with 1.15 Sharpe Ratio Track Record

Core Thesis

Morgan Stanley's Earnings Surprise Composite model, which synthesizes signals from forecast landscape, earnings quality, and broader forecast dynamics, identifies stocks with the highest probability of exceeding quarterly expectations. When combined with fundamental analysts' Overweight ratings, this quantamental approach has delivered pre-cost Sharpe ratios of 1.15 (US) and 0.94 (Europe) since live tracking began in 2024. Four US stocks—Applied Materials (AMAT), Intuit (INTU), CrowdStrike (CRWD), and Nvidia (NVDA)—carry the highest conviction among top-quintile candidates reporting between May 7 and June 2, 2026.

Evidence Chain

1. The composite signal systematically outperforms single-factor approaches. The model aggregates three quasi-independent dimensions. Earnings Forecast Landscape captures trend and breadth of analyst revisions. Earnings Quality assesses accruals, cash flow divergence, and sustainability metrics. Broader Forecast Dynamics measures macro or sector-level adjustments that may not yet flow into company-specific estimates. Stocks ranking in the top quintile on all three legs generate substantially higher hit rates than those ranking high on only one factor. Since 2019, the monthly hit rate has been 62.9% for US stocks and 60.0% for European stocks—meaning the top-quintile portfolio beat expectations more than three out of five months over a seven-year period.

2. Live out-of-sample performance exceeds backtest metrics. The strategy's CAGR/Vol ratio improved from 0.72 (2019–2023 backtest, US) to 1.15 (2024–2026 live, US). European figures show a similar trajectory—0.71 to 0.94. Maximum drawdown narrowed from -12.4% to -8.1% (US) and from -11.9% to -8.6% (Europe). The Calmar ratio doubled in both regions to 1.30. This improvement may reflect the model's ability to capture regime shifts in analyst behavior post-pandemic, though two years of live data remain insufficient for definitive attribution.

3. Current highest-conviction US candidates cluster in tech but span different score profiles.

• Applied Materials (Composite: 0.87, report date: May 13): Top decile on Forecast Landscape (0.85) and Broader Forecast Dynamics (0.88), with moderate Earnings Quality (0.41). The model favors this name despite weak quality metrics, suggesting analysts are actively raising estimates ahead of a capital equipment cycle inflection.
• Intuit (Composite: 0.82, report date: May 19): Balanced across Forecast Landscape (0.51), Earnings Quality (0.79), and Broader Dynamics (0.69). High quality score reduces the risk of accrual-related disappointment.
• CrowdStrike (Composite: 0.70, report date: June 1): High Earnings Quality (0.75) and moderate Forecast Landscape (0.44), indicating the surprise potential derives from operational execution rather than analyst excitement.
• Nvidia (Composite: 0.69, report date: May 19): Broader Forecast Dynamics at 0.98—the highest in the US cohort—but Earnings Quality at 0.01, the lowest. This extreme divergence suggests the model captures massive upward estimate momentum, but investors should anticipate higher volatility in the quality dimension.

4. European top picks show stronger composite scores with geographic concentration. ABN AMRO Bank (Composite: 0.99) leads all names globally, with near-perfect scores across all three subcomponents. Other top European names include Assicurazioni Generali (0.93), International Consolidated Airlines Group (0.88), and Ferrovial SE (0.72). The European cohort is more diversified by sector—financials, industrials, and materials—versus the US cohort's heavy tech tilt.

5. Low-scoring stocks present identifiable short candidates. The model also identifies stocks with composite scores in the bottom quintile. US examples include Heico (0.01), Hewlett Packard Enterprise (0.03), and Dell Technologies (0.06). European examples include Imperial Brands (0.02), Coloplast (0.03), and Alstom (0.05). These names share weak Forecast Landscape signals and poor Broader Forecast Dynamics, implying that downward estimate revision momentum is already embedded.

Investment Implications

The strategy is designed as a cross-sectional relative-value trade, not a directional long portfolio. The evidence supports a long/short approach: long the top-quintile Overweight-rated names and short the bottom-quintile Underweight/Equal-weight names. The 64.7% monthly hit rate on the long side (US live) provides a high probability of positive contribution from the long book, while the short side captures names where estimates are likely to decline further.

For investors who cannot short, concentrating long positions in names with balanced subcomponent scores—such as Intuit or Applied Materials—reduces the risk of factor-specific disappointment. Nvidia's extreme scoring divergence warrants caution: the model strongly predicts an earnings beat driven by revenue momentum, but the near-zero Earnings Quality score means any deviation from the demand narrative could trigger a disproportionate penalty.

Key Risks

1. Model regime change. The composite's performance improvement from backtest to live may reflect a period favorable to momentum-oriented signals. A shift to a regime where quality or valuation dominates—such as a sudden tightening of financial conditions—could disrupt the forecast dynamics leg.

2. Sector concentration in US. Eight of the top 16 US names are in Information Technology. Any sector-specific shock (e.g., export controls, cloud spending deceleration) would hit the long book disproportionately.

3. Trading frequency and cost. The strategy trades monthly only during earnings months (May, then July refresh). Transaction costs and market impact are not netted—pre-cost Sharpe ratios overstate net returns, especially for the short book where borrowing costs can be material for names like Heico or Imperial Brands.

4. Single-factor failure mode. The model does not incorporate valuation, balance sheet leverage, or macro hedging. A stock with high composite scores but extreme valuation—such as Nvidia at a $4.8 trillion market cap—could beat earnings yet still decline if expectations are already too high.

Appendix Data Summary

Key Performance Metrics (Pre-Cost)