Dollar ETF vs Gold ETF

UUP is the Invesco DB US Dollar Index Bullish Fund, launched February 20, 2007. It holds long futures positions in the ICE Dollar Index (DXY): euro 57.6 percent, JPY 13.6 percent, GBP 11.9 percent, CAD 9.1 percent, SEK 4.2 percent, CHF 3.6 percent. UUP rebalances daily to track the index. AUM was approximately 425 million as of April 2026, with average daily volume around 1 to 2 million shares. The expense ratio is 0.78 percent, high relative to broad equity ETFs but typical for futures-based commodity and currency products. UUP carries K-1 tax reporting because it is structured as a commodity pool.

GLD is the SPDR Gold Shares ETF, launched November 18, 2004. It holds physical gold bullion in HSBC vaults in London, with shares representing 0.09405 ounces of gold each as of April 2026 due to the cumulative 0.40 percent annual expense ratio. GLD AUM was approximately 105 billion as of April 2026, the largest gold ETF globally and the largest commodity ETF by AUM. Average daily volume runs 8 to 12 million shares. GLD does not generate K-1s but carries collectibles tax treatment for long-term holders (28 percent maximum federal rate).

Rolling correlation analysis shows the UUP-GLD relationship is consistently negative but variable. Across the full 2007 to 2026 sample, the correlation in monthly returns averages minus 0.46. On 90-day rolling daily-return windows, the correlation averages minus 0.44, ranges from minus 0.81 (most-inverse, May 2010 European debt crisis window) to plus 0.32 (most-positive, July 2022 simultaneous safe-haven rally), and spends roughly 78 percent of windows in negative territory.

On 1-year windows (260 trading days), correlation averages minus 0.55 and ranges from minus 0.78 to minus 0.05; the relationship is more reliable at this horizon. On 5-year windows, correlation averages minus 0.62, with the lowest 5-year reading at minus 0.78 (window ending 2014, capturing the dollar-strength and gold-weakness regime). The April 2026 90-day rolling correlation reads minus 0.51, slightly more negative than the long-run average and consistent with the mechanical pricing channel dominating over idiosyncratic factors during the current Fed-cut and Iran-tail-risk environment.

Two channels link UUP to GLD. First, the priced-in-dollars mechanical channel. Gold trades globally in US dollars; when the dollar appreciates, the same ounce of gold becomes more expensive in EUR, JPY, GBP, and INR terms, dampening physical demand and weighing on the dollar gold price. The mechanical effect is roughly 0.3 to 0.5 percent gold price decline per 1 percent dollar appreciation, with lag of zero to ten trading days depending on market liquidity.

Second, the real-rate channel. Dollar strength typically reflects either Fed hawkishness (which raises US real yields and the opportunity cost of holding non-yielding gold) or risk-off flows (which can support both currencies depending on the shock). The real-rate effect is larger and slower, typically 0.5 to 1.0 percent gold price change per 25 basis point real-yield move with a lag of 2 to 6 weeks. The April 2026 environment shows both channels contributing: the dollar has weakened on Fed cuts and reduced US-rate premium versus Europe and Canada, and 10-year TIPS yields have fallen from 2.10 percent in late 2024 to 1.85 percent in late April 2026, both consistent with the gold rally.

Three named episodes broke the UUP-GLD inverse link materially. First, the 2008 GFC simultaneous safe-haven rally (September to November 2008): UUP rose 13 percent and GLD fell only 5 percent (versus expected 7 to 10 percent gold decline given the dollar move) because central-bank intervention drove the dollar bid and gold hedging demand simultaneously. The 90-day rolling correlation rose to minus 0.05 by November 2008.

Second, the 2014 to 2015 dollar-strength gold-weakness regime: UUP rose 22 percent from June 2014 to March 2015 while GLD fell 9 percent, more than the mechanical channel predicted. The amplifier was that ECB QE was launched and the BoJ doubled-down on QQE, creating both currency-side dollar strength and additional gold weakness. The 90-day correlation reached minus 0.78, the most negative in the sample, but the level shift was structural rather than typical.

Third, the August 2022 simultaneous rally: UUP rose 4 percent and GLD rose 2 percent over the August 2022 window as dollar safe-haven flows on Russian-gas concerns coincided with gold safe-haven demand. The correlation went briefly positive (plus 0.32) in July to August 2022. Each of these regime breaks reset the level around which the spread oscillated for several months before reverting.

UUP closed at 27.10 on April 30, 2026, down approximately 7 percent year-to-date. The DXY closed at 98.56 on April 24, 2026, near the early-March low of 98.49, with 55-day moving average providing support. GLD closed at 472.50, up approximately 36 percent year-to-date. Spot gold has retraced to around 4,720 per ounce after peaking at 5,602.22 on January 28, 2026.

The configuration is the textbook inverse setup: dollar weakening, gold strengthening. The 90-day rolling UUP-GLD correlation is minus 0.51, slightly more negative than the long-run average. The primary drivers: Fed cuts since September 2024 (100 basis points cumulative through December 2024) compressed the rate differential; central-bank gold buying continued at 244 tonnes in Q1 2026 (World Gold Council Gold Demand Trends Q1 2026); Iran-related geopolitical risk premium added an estimated 200 to 400 dollars per ounce to gold from late February 2026 forward. The correlation could compress further if Iran shock subsides and the Fed pauses cuts, but the structural inverse anchor remains intact.

Three implementation styles dominate. First, pair trades: long GLD, short UUP, sized for a target spread vol of 8 to 12 percent annualized. The trade captures the inverse relationship during dollar moves but loses when both legs move together (like 2008 or July 2022). Two-thirds of pair trades historically profit; the average winner is approximately 4 to 6 percent over a 90-day holding window, the average loser is approximately 3 to 4 percent.

Second, hedging overlays: international equity allocators use UUP as a partial dollar hedge, GLD as a partial real-rate hedge, with both legs offsetting fiat-debasement risk. The overlay typically sizes at 5 to 10 percent of total portfolio AUM. Third, options-based volatility trades: GLD implied volatility in April 2026 averaged 18 to 22 percent versus realized of 15 percent, while UUP implied averaged 9 to 11 percent versus realized of 8 percent. The implied-realized gap on GLD has been wider than UUP throughout 2026, reflecting concentrated demand for gold tail-risk protection. Long GLD straddles versus short UUP straddles is one expression of expected divergence in vol regimes.