T48: Intraday GARCH VT — leverage artifact, not improvement

G28/I1 showed intraday GARCH improves QLIKE by +10.3% for vol prediction. Tested whether this translates to better VT strategy. Intraday GARCH VT Sharpe=1.36 vs Total GARCH VT 1.06 (DM p=0.005, bootstrap p=0.011). BUT: this is accidental leverage — intraday sigma is 83% of total sigma, so weights are systematically +16.6% higher. In 2023-2024 bull market this = higher returns. Evidence: (1) QLIKE is WORSE (+7.8%) for predicting total return vol, (2) MDD is worse (-11.7% vs -10.3%), (3) up-day gain +11.1 bps exactly offset by down-day loss -9.6 bps, (4) gap adjustment only triggers 5% of days — negligible. Verdict: NULL. Intraday GARCH improves vol prediction quality but the VT benefit is a leverage artifact, not genuine risk-adjusted improvement.