This paper extends the Bitcoin Media Pressure Index (BMPI) developed in Parfenovich (2026, SSRN Working Paper No. 6542901)-a daily media-sentiment signal derived from GDELT Global Knowledge Graph 2.0-from a validated signal into a systematic portfolio construction tool. Using 3,800 daily observations spanning September 2015 through January 2026, we demonstrate that BMPI, which measures the information environment around Bitcoin through normalised mention volume and coverage tone, can govern dynamic exposure in a way that improves risk-adjusted performance. A cascade allocation strategy (V9, SHARPE_MODE) that scales position size according to BMPI-defined regimes-supplemented by volatility targeting and drawdown protection-achieves an out-of-sample Sharpe ratio of 2.27 and a maximum drawdown of-16.6% on a 1,120-day test window (December 2022-January 2026), versus 1.38 and-32.6% for passive buy-and-hold. We do not claim that BMPI forecasts return direction; the mechanism is exposure management, not market timing. The Jensen alpha is +0.418 per annum (t = 3.864, p < 0.001, HAC-corrected). A permutation test comparing BMPI-weighted against fixed-weight allocation yields p = 0.029. In a live-deployment scenario with expanding-window BMPI calibration, the strategy retains its performance characteristics. BMPI-based allocation survives doubled transaction costs and two-day execution latency. Honest limitations are discussed, including the non-significance of the trailing-stop contribution under permutation and the look-ahead risk in full-sample calibration.