A rising dollar is a headwind for most equities, but brief DXY spikes are noise. This composite gate keeps only sustained strength: DXY above 100, 15-day momentum positive, and price above its 65-day average. All three must hold. That fires on just ~10% of days since 2009 — a rare, high-conviction tail-risk regime, not an everyday signal. As of August 31, 2026 the signal is OFF (DXY 99.4).

Sector returns are mildly negatively correlated with the dollar day-to-day. The edge is that when DXY clears 100, that inverse relationship intensifies for rate-sensitive sectors. Real Estate (XLRE) is the extreme — its correlation amplifies by -0.25, the most of any sector — followed by Comm Services and the defensives. Energy is the lone inversion: its correlation turns less negative (+0.053), because dollar-denominated commodity contracts rally with the macro backdrop that drives the dollar up.

At the one-month horizon the regimes pull apart with statistical significance. Energy (XLE): +2.8% in strong-$ vs +0.8% in weak-$, a +2.0% edge (***, p=0.0000) — the cleanest signal in the study and a near-exact match to the source deck. Financials lean positive (steeper curves), while the rate-sensitive defensives (Staples, Utilities, Real Estate) underperform in strong-$ windows. The edge is a 21-day phenomenon — too noisy at 1 day, matured by a month.

• XLE — The Structural Hedge. The only sector that benefits from strong $ (amplification +0.053, +2.0% 21d edge). Overweight during DXY>100 to capture commodity rallies — do not default to cash here (filtering XLE historically worsens drawdown by 1pp and costs the biggest rallies).
• XLP — The Alpha Generator. A weak-dollar compounding engine; mild strong-$ headwind, and filtering strong-$ periods improves its drawdown (+4pp). The core defensive to rotate back into when the signal turns off.
• XLRE — The Rate Reactor. The most regime-reactive sector (-0.25), highly binary. Strictly avoid when the composite signal fires.
• XLC — The Drawdown Trap. Neutral on a single day but devastating in the tail — filtering strong-$ windows avoids +18pp of max drawdown (the 2022 multiple-compression slaughter). A strategic-avoidance candidate for tail-risk control.

When the signal is ACTIVE (now): overweight XLE · XLF · XLI · SMH and reduce/avoid XLRE · XLP · XLU · XLC. When it turns OFF: rotate back into the rate-relief beneficiaries (Staples, Real Estate, Utilities, Comm Services, Discretionary). The signal onset historically precedes the sector rotation by 5–21 days, so a fresh firing is a prepare signal, not a same-day scramble.

• Tail-risk overlay, not an engine. It is active only ~10% of the time — use it to tilt a portfolio at the margin, not to build one.
• Keep Energy exposure. The most common mistake is defaulting to cash on a strong-$ signal; for XLE that destroys value. Express the macro tilt by overweighting XLE, not exiting it.
• Honest replication notes. The core (XLE +2.0% / p<0.001) reproduces the deck closely. Some sectors are less statistically significant on yfinance data than the deck reported, and XLRE (2015+) and XLC (2018+) have shorter histories than the headline 2009 window. The 2022 validation (signal active 41.8% of that year) is the cleanest real-world stress test. In-sample, not a walk-forward backtest; transaction costs unmodeled.