Active Manager Holdings Are Persistent Alpha Signals, Not Just Positioning Data

Core Thesis

US active mutual fund disclosures contain systematic, investable equity selection signals that survive rigorous risk adjustment. Nine distinct signals built from 13F-derived holdings changes all deliver positive risk-adjusted returns over 2016–2026. The top performer, Mean Holding Weight, produces a 17.5% CAGR at a 0.92 Sharpe—translating to annualized alpha of 2.5% to 4.0% after standard equity factor controls. This implies that institutional positioning information is priced gradually enough to leave a window for systematic capture.

The Dimension Investors Often Miss

The market treats most 13F filings as stale, low-resolution snapshots. That consensus underweights how aggregation solves the noise problem. When stock-level scores distill position changes across hundreds of funds over a rolling 45-day lookback, the resulting signal isolates coordinated institutional conviction rather than idiosyncratic manager noise. The disclosure lag does not neutralize the information; it merely stretches the repricing horizon.

Three signal construction families prove this point:

• Mean Holding Weight captures the average portfolio weight managers assign to a stock—a direct gauge of conviction that scales with position size.
• Herding Score identifies stocks where multiple managers are simultaneously increasing exposure, capturing coordinated accumulation before consensus forms.
• Reallocation Intensity measures the speed and magnitude of cross-sectional weight shifts, producing an Information Ratio of 0.76 against the S&P 500—substantial in long-only equity.

The fact that Sharpe ratios survive a one-third reduction across sub-periods (2016–2019 vs. 2020–2026) argues against data mining. Parameter stability across lookback windows further supports economic substance over curve-fitting.

Evidence Chain

Nine long-only, quarterly-rebalanced portfolios (top 30 stocks, 45-day lookback) produce CAGRs between 14% and 19%. Sharpe ratios range from 0.62 to 0.92. The top three signals all exceed 0.88 and pass the Benjamini-Hochberg multiple-testing correction at a 5% false discovery rate.

Risk-factor attribution is the critical filter: annualized alpha of 2.5% to 4.0% survives controls for size, value, momentum, and other standard equity factors. This separates genuine information content from factor repackaging. A max-diversification ensemble of three signals delivers the highest Sharpe in the study, confirming that signal construction methods pick up complementary aspects of institutional behavior.

Regime analysis reveals a concentrated vulnerability: most signals remain positive across VIX quartiles and bull/bear conditions, but stress concentrates in the High VIX / Bear cell. This aligns with the mechanism—forced deleveraging in risk-off environments disrupts the gradual repricing that these signals exploit.

Key Risks

The vulnerability cluster in high-volatility, bear-market regimes is structural. When cross-asset correlations spike and redemptions force indiscriminate selling, the informational content of active manager holdings degrades. Position changes in those environments reflect liquidity constraints, not conviction. Any systematic strategy built on these signals requires explicit drawdown management during VIX spikes.

Lookback parameter tuning introduces a model risk that performance summaries can obscure. While Sharpe ratios are broadly stable across the grid, live implementation involves real-time decisions about decay rates and rebalancing frequency that interact with transaction costs—a friction absent from this simulation.

The research presents a hypothetical illustration of mathematical principles. No actual fund track record exists. Implementation costs, market impact from crowding into manager-favored names, and capacity constraints are not modeled.

Trade Implications

For systematic equity allocators, these signals function as standalone alpha sources or portfolio overlays. The 0.76 Information Ratio from Reallocation Intensity is competitive with many commercial factor products. An ensemble approach using the max-diversification weighting scheme offers the strongest risk-adjusted profile while reducing single-signal regime sensitivity. The stress-concentration pattern in high-VIX bears suggests pairing these signals with a volatility-contingent sizing rule: reduce gross exposure when the VIX crosses into its top quartile, rather than treating allocation as static across all environments.