Two new analytical pieces challenge popular Bitcoin narratives about central bank reserve diversification and stock-market decoupling. New York Fed researchers Linda S. Goldberg, Oliver Hannaoui and Sneha Parthasarathy show that the dollar’s falling share of global official foreign-exchange reserves does not by itself signal a broad retreat from dollars or rising sovereign Bitcoin demand. Using IMF COFER data, they report the dollar share fell from 64% at year-end 2015 to 56% at year-end 2025.
The researchers separate reserve changes into a “preferences” channel — countries altering their currency mix — and a reserve-size channel, where changes in a country’s total reserves alter its weight in the global average. Switzerland is cited as a key example: between 2015 and 2019, its reserve growth pulled down the aggregate dollar share even as its own dollar allocation rose. For 2015–2019, among 76 countries with complete data, preferences and reserve-size changes reduced the dollar share by 1.2 and 1.5 percentage points respectively. For 2019–2023, the 62 countries with complete data contributed a positive 0.3 percentage point through preferences and a negative 0.5 percentage point through reserve-size changes. China, Russia, Mexico and Morocco lacked 2023 dollar-allocation data; the researchers inferred a combined negative 2.0-percentage-point preferences contribution for that group under assumptions matching the observed 2.3-point global decline.
The underlying Staff Report 1087, issued in March 2024 and revised in February 2026, models reserves as consisting of a liquidity portion and an investment portion above those needs. Diversification is more prevalent when reserves exceed liquidity needs, but the authors stress that diversification does not reveal the destination of funds. On Nov. 13, 2025, the Czech National Bank announced a $1 million digital-asset test portfolio including Bitcoin, a dollar stablecoin and a tokenized deposit, with the purchase explicitly outside international reserves. The Fed research does not measure sovereign Bitcoin purchases or estimate a Bitcoin price effect. A sovereign Bitcoin demand case would need separate evidence including a disclosed allocation, funding source and executed purchases.
A separate Bitwise Market Compass published Sept. 7 notes Bitcoin’s 260-day correlation with the S&P 500 is the lowest since 2015. But CryptoSlate’s calculations using FRED data show the correlation depends heavily on methodology. The negative reading tracks logarithms of price levels and was roughly −0.62, while the correlation of percentage returns over 260 common trading days was +0.40. Because FRED records Bitcoin at 5 p.m. PST and the S&P 500 at the U.S. market close, the two series share a calendar date but not a timestamp, limiting precise conclusions.
When equities fell on 115 common dates in the 260-return sample, Bitcoin fell on 76 of them and averaged a 1.03% decline. On the 25 days the S&P 500 fell more than 1%, Bitcoin fell on 22 and averaged a 2.59% loss. In a hypothetical portfolio replacing 5% of an equity allocation with Bitcoin, measured annualized volatility rose from 12.68% to 13.01% over the 260-return window and maximum drawdown increased from 9.10% to 9.98%. Over a 500-return window, the same allocation modestly reduced volatility and drawdown, showing that conclusions change with period and measurement. Bitwise itself stops short of treating the decoupling as a hedge, warning that decouplings usually do not last.
The combined takeaway for investors is caution: falling dollar reserve share is not evidence of central bank Bitcoin buying, and a low log-level equity correlation does not mean Bitcoin will offset stock losses on bad days.
