ISOLATING MARKET TIMING ALPHA THROUGH CONTROLLED FACTOR EXPOSURE: A Preregistered Validation Methodology Using Fama-French, Treynor-Mazuy, and Henriksson-Merton Tests

This preregistration presents a novel experimental design for isolating market timing alpha from factor exposure. Standard market timing research suffers from a fundamental confound: timed portfolios typically hold different securities than untimed benchmarks, making it impossible to attribute performance differences to timing versus selection. We address this by constructing controlled experiments where factor exposures are identical by construction between timed and untimed portfolios, such that any improvement in factor-adjusted metrics must be attributable to timing. We test three studies using established econometric frameworks: Fama-French six-factor model, Treynor-Mazuy quadratic regression, and Henriksson-Merton asymmetric response model. Study 1 tests exit timing using benchmark strategies with zero proprietary entry signals. Study 2 tests entry timing using Temporal Structural Forecasting (TSF) signals with native factor exits. Study 3 tests full TSF timing (entry and exit) against native factor portfolios. The analysis covers 346 S&P 500 stocks (those current constituents with sufficient historical data to generate 20 years of TSF forecasts) across 6 factors, 3 universes, 3 position sizes, 3 confidence bands, and 4 time periods (2006-2025), yielding 2,392 individual comparisons. Sixteen hypotheses are preregistered across three econometric models. For each hypothesis, we report improvement rates tested against a 50% null using binomial test. This methodology establishes a rigorous framework for evaluating market timing claims while controlling for factor exposure confounds.