Diversification is often described as a fixed property of a portfolio, but the correlation between asset classes is not static. It moves in regimes, sometimes for years at a time, and the current environment offers a useful case study in how those regimes show up in the data before anyone assigns them a label.
The 60/40 correlation question
The rolling 60-day correlation between the S&P 500 and the 10-year Treasury has been the most closely watched cross-asset relationship since 2022, when it flipped from its long-standing negative range of roughly -0.3 to -0.5 into positive territory above 0.5 for extended stretches. As of the most recent data, that correlation sits near 0.35, down from the extremes of 2022 but still meaningfully above the 2010-2021 average. This matters because the negative correlation of that earlier period was the mechanical reason a 60/40 portfolio dampened drawdowns: when equities fell, bonds tended to rise, offsetting some of the loss. A positive correlation regime removes that offset during periods when both asset classes are repricing off the same input, typically shifts in the expected path of policy rates.
What the last twelve months show
Over the trailing twelve months, there have been three distinct windows where stocks and bonds moved together on the same days rather than against each other. Each of these windows lined up with a repricing of rate expectations rather than a growth scare, which is consistent with the post-2022 pattern: when inflation or policy rate expectations are the dominant driver of markets, stocks and bonds behave like substitutes reacting to the same discount rate. When growth or credit-quality concerns dominate instead, the older negative correlation tends to reassert itself, as it did briefly in spring 2025 during a short-lived growth scare when 10-year yields fell roughly 40 basis points while the S&P 500 declined more than 6% over a two-week span.
Gold's parallel move
Gold has behaved somewhat differently from the textbook description of a pure inflation hedge. Its correlation to real yields, historically negative and fairly tight, has loosened over the past eighteen months, with gold registering gains even during stretches when real yields were flat to modestly higher. Central bank purchasing activity, which the World Gold Council has estimated at over 1,000 tonnes annually in recent years compared with an average closer to 500 tonnes in the decade prior, appears to be a structural source of demand that is not purely a function of the U.S. real rate cycle. This is a useful reminder that a single asset's correlation to a traditional driver can shift when the composition of buyers changes, independent of anything happening in equity or bond markets.
Dollar and international divergence
The U.S. dollar index has spent much of the past year in a relatively narrow band compared with the wide swings of 2022, when the DXY moved from the mid-90s to above 114. That reduced volatility in the dollar has coincided with a narrower performance gap between U.S. and international developed equities than was typical during the dollar's steep 2021-2022 ascent, when a strong dollar was a persistent headwind for unhedged international returns. Currency has historically been one of the larger unpriced factors in international allocations, and periods of dollar stability tend to reduce the noise between local-currency and dollar-based return figures, making the underlying equity performance easier to read in isolation.
Volatility of correlation itself
One detail that gets less attention than the correlation level itself is the volatility of that correlation over time. The standard deviation of the 60-day stock-bond correlation has been elevated relative to its own longer-run history, meaning the relationship has been swinging between positive and near-zero readings more frequently than in prior decades. Data going back to the early 1990s shows that correlation regimes, once established, have tended to persist for two to four years on average, but transition periods, like the one that appears to be underway now, are typically choppier and less predictable than either the stable positive or stable negative regimes that precede and follow them.
Why regime matters more than level
The single correlation number reported on any given day is less informative than the regime it belongs to. A reading of 0.3 during a transition period carries different implications than the same reading during a stable multi-year regime, because the drivers behind it are less settled. Historically, the transition from a negative to a positive stock-bond correlation regime, or vice versa, has coincided with a shift in the dominant macro variable, from growth-driven pricing to rate-driven pricing or back again. Recognising which variable is currently doing the work has tended to be a more useful exercise than tracking the correlation coefficient in isolation.
The rule to internalise
Correlation between asset classes is a description of a moment in the macro cycle, not a fixed engineering constant of a portfolio. The historical value of a diversified allocation across stocks, bonds, gold, and currencies has come less from any one correlation holding at all times and more from the fact that these relationships have shifted enough, across enough different periods, that no single pairing has stayed broken indefinitely. Reading today's correlation snapshot as one data point within that longer, shifting pattern, rather than as a permanent feature of the current market, is the more durable way to interpret it.
Educational content only. Not investment advice.