Cross-Asset Correlation: Why Stocks, Bonds, Gold, the Dollar and Bitcoin Move Together or Apart

What this guide covers

What correlation is and why diversification comes from it were covered in the guide on diversification and correlation. This guide widens the view to very different assets, stocks, bonds, gold, the dollar and bitcoin, and how they are linked. Stocks within one market tend to move for similar reasons, but different asset classes move for different ones. Even so, there are periods when they rise or fall together and periods when they move in opposite directions. What makes the difference is shared drivers that affect several assets at once. Below are the main shared drivers, followed by the traps that distort correlation when you measure it yourself.

Shared drivers that move assets together

The forces that affect many asset prices at the same time can be summarized in a few groups. Relationships between assets change depending on which driver dominates a given period. The descriptions below are the commonly used general structure; how well they held varies by period, so check with data.

• Interest rates: bond prices move opposite to rates. Higher rates reduce the present value of future profits, which is said to weigh on stocks, especially technology stocks that depend on distant growth
• Risk appetite: when people are willing to take risk, risk assets such as stocks and bitcoin tend to rise together, and when they are afraid, to fall together
• The dollar: a stronger dollar tends to weigh on gold and commodities priced in dollars and on the dollar value of non-dollar assets
• Inflation: when inflation is the main worry, rising rates can weaken stocks and bonds at the same time
• Liquidity: as money is added to or drained from the system, many risk assets often move the same way

Stocks and bonds: the sign can flip

Stocks and bonds are the classic diversification pair, but their correlation is not fixed. When slowing growth is the main worry, rates fall as stocks drop, so bond prices rise and the two tend to move in opposite directions. When rising inflation is the main worry, rates go up and stocks and bonds can fall together. A widely known example is 2022, when inflation surged, rates rose quickly and US stocks and bonds both fell sharply. So 'bonds rise when stocks fall' is an observation about certain periods, not a law. When reading correlation, look at how the value changed across periods rather than at a single full-period number.

Gold, the dollar and bitcoin

Gold pays no interest or dividends, so higher rates raise the opportunity cost of holding it, and since it is priced in dollars, it is commonly said to be affected by the dollar's strength. Demand for gold can also rise when serious anxiety spreads. The dollar is often said to strengthen when global markets are shaken, which matters especially to investors in Korean won: if the dollar-won rate rises in a risk-off phase, part of the loss on overseas assets measured in won is offset. Bitcoin trades around the clock and has a short history, so its relationships are less stable. In some periods it moves much like technology stocks, and in others it moves on its own. Rather than labelling bitcoin as either 'digital gold' or 'a risk asset like tech stocks', it is better to measure, period by period, which it currently moves more closely with.

Correlation is not fixed: rolling correlation

A single correlation figure is an average over the whole period you calculated. The same two assets can give very different values depending on the period, so also look at rolling correlation, recalculated each day over a fixed window such as the last 30 or 60 days. If rolling correlation hovers around zero there is no clear relationship; if it stays on one side for a long time the relationship held in that period. The shorter the window, the fewer the samples and the more the value swings, so do not read a sudden move in a short window as a change in the relationship right away. The figure below is an example made to show how a correlation figure arises.

Correlation is not size

Correlation only says how often the direction matched, not how large the moves were. Even with a correlation of 1, one asset might move 3% when the other moves 1%. To capture size, use beta. Beta is covariance ÷ variance of the reference asset, which equals correlation × (the asset's volatility ÷ the reference asset's volatility). So a very volatile asset can have a large beta even with a moderate correlation, and a highly correlated asset with low volatility can have a small beta. When thinking about how much a whole portfolio swings, correlation, each asset's volatility and the weights all work together. An asset with low correlation may not deliver the expected effect if its weight is tiny or its volatility very large.

Traps when measuring correlation

Correlation is simple to compute, but the number changes a lot depending on what you feed in and how. These problems come up especially often when mixing asset classes. If a correlation you read elsewhere differs from yours, one of these is usually the cause.

• Correlating price levels gives high values whenever both simply rise. Always use returns
• Interest rates are not prices, so measure them as point changes, not percentage changes
• Markets that close at different times (KOSPI and US indices) cover different hours under the same date, which tends to lower correlation. Weekly returns reduce the effect
• Crypto trades on weekends, so when matched with stocks, weekend moves pile into Monday's change
• Small samples (short periods) easily produce values that are high or low by chance
• Measuring in won and in local currency gives different values because the exchange rate enters

Using the tools on this site

This site's Bitcoin vs Stocks, Gold and Dollar Correlation tool compares bitcoin or ether with the Nasdaq, S&P 500, KOSPI, gold, the dollar index, the US 10-year yield, WTI crude and the dollar-won rate, showing 30-, 90-, 180- and 365-day correlations as bars and a matrix, and 30- or 60-change rolling correlation with a chosen asset as a chart. Unfinished candles are not used, and the rate is measured as point changes. In Stock Compare you can pick up to five Korean or foreign stocks, ETFs, indices, gold or currencies and see a return correlation matrix alongside beta, annual volatility and maximum drawdown, choosing daily or weekly returns for correlation. For how exchange rates affect returns, see the guide on currency effects on overseas stocks.

Limits and disclaimer

Correlation summarizes how much assets moved together in a past period; it does not explain causes and does not guarantee the relationship will continue. Relationships often look different from usual when markets are turbulent. The shared-driver descriptions here are a commonly used general structure and do not fit every period. The figure is an illustration of the principle, not a record of real assets. This guide explains how to read relationships between assets, does not recommend any asset allocation and is not investment advice.