ETH/BTC Near 0.03: What Copper/Gold Actually Tells Us

The macro intuition is reasonable. Copper is widely used in construction, manufacturing, power grids and infrastructure, making demand sensitive to expectations for growth and industrial activity. Gold plays a more defensive role and tends to attract demand when investors are seeking protection. When copper outperforms gold, markets may therefore be pricing stronger economic activity and a greater preference for growth-sensitive assets.

A comparable framework can be applied to crypto markets. Bitcoin (BTC) retains a relatively more monetary profile within digital assets, while Ethereum is more exposed to network activity, blockspace demand and the development of decentralised applications. During periods of stronger risk appetite and expanding crypto activity, Ether (ETH) may therefore benefit more than bitcoin from the broader environment.

A plausible macro narrative, however, does not automatically translate into a reliable predictive signal. Our analysis finds an association under certain conditions, but one that is heavily concentrated around the 2020–2021 episode.

(Source: TradingView)

The macro case for the comparison

Copper/gold is often viewed as a measure of the balance between growth and caution. A rising ratio can reflect improving industrial expectations, stronger demand for cyclical assets, or a reduced preference for safe havens. A falling ratio can point toward weaker growth expectations and increasing demand for protection.

That makes the comparison with ETH/BTC intuitive. In an environment of expanding risk appetite, crypto assets with greater sensitivity to activity within the sector may outperform assets with a more monetary profile. Ethereum may benefit from its stronger exposure to onchain activity, although this does not imply that copper/gold causes ETH/BTC to rise or that the relationship can be used systematically for market timing.

Both ratios may simply respond to common forces. Global liquidity, the dollar, real interest rates and broader risk appetite can influence industrial commodities and crypto markets at the same time. The key question is therefore not whether the two ratios sometimes move in a similar direction, but whether copper/gold contains stable information about future ETH/BTC returns.

How the test was designed

The analysis covers January 2017 to July 2026. Copper/gold is first constructed using copper and gold prices from trading platform OANDA, before the main tests are repeated with continuous COMEX futures to assess whether the conclusions depend on the underlying data source.

The primary specification was defined before examining the results. It measures the combined effect of weekly copper/gold returns observed between 13 and 26 weeks earlier on current ETH/BTC returns. Returns are used instead of price levels to reduce the risk of identifying artificial relationships between assets that merely share a long-term trend.

Newey-West standard errors are applied because financial series can display changing volatility and correlation between nearby observations, particularly when return periods overlap. The US Dollar Index, the United States ten-year real yield and the bitcoin market regime are also included as controls, so that any apparent link is not simply the two ratios responding to the same macro backdrop. An out-of-sample exercise then evaluates the model from 2022 onward using data that were not part of the original estimation.

The primary test finds nothing

Across the full sample, the distributed-lag model, which measures the combined influence of a run of earlier periods rather than a single one, does not identify a stable positive cumulative effect from copper/gold to ETH/BTC between 13 and 26 weeks.

The estimated cumulative effect is −0.10 with a p-value of 0.855, meaning a result at least this strong would arise by chance most of the time even if no relationship existed. After controlling for the dollar, the United States ten-year real yield and the bitcoin regime, the estimate falls to −0.26 with a p-value of 0.654. Both results are far from the conventional significance threshold of 0.05.

Individual weekly coefficients also alternate between positive and negative values, rather than forming a coherent positive sequence. In the COMEX robustness test, one isolated negative lag remains significant after adjustment, but this does not support the idea that stronger copper/gold consistently precedes stronger ETH/BTC. Taken together, the primary test provides no evidence that a rise in copper/gold is systematically followed by ether outperforming bitcoin.

Across different horizons

Direct regressions provide a second way to examine the relationship. Each specification compares copper/gold momentum over a fixed period with ETH/BTC returns during the following weeks. These tests are exploratory. They extend the analysis beyond the primary specification and can reveal potentially useful patterns, but they provide weaker evidence because the additional combinations are examined after the main test has been defined.

All four beta coefficients, which measure the estimated size of the relationship, are positive, meaning that stronger copper/gold momentum is associated with higher subsequent ETH/BTC returns in these models. The two most interesting specifications use copper/gold momentum over 26 weeks.

In the final specification, a 1% increase in copper/gold is associated on average with an approximately 0.82% increase in ETH/BTC over the following 26 weeks. Its raw p-value is 0.043, while the 26-week to 13-week specification has a raw p-value of 0.047.

Both results initially meet the conventional 5% threshold. Once the Benjamini-Hochberg adjustment, which raises the bar for significance when several tests are run at once, accounts for the four specifications tested simultaneously, the two p-values rise to approximately 0.095. They therefore no longer meet the 5% threshold.

The strongest specification explains around 10% of subsequent ETH/BTC variation within the sample. That is economically intriguing, but it does not establish that the model would have generated reliable real-time forecasts. The 26-week future return windows also overlap heavily, reducing the amount of genuinely independent information contained in the sample.

Almost all of it sits in one episode

A comparison across subperiods shows how unstable the relationship has been. It remains weak before 2021, becomes exceptionally strong around 2021 and fades substantially after that episode.

Within the 2020–2021 window, some direct regressions explain as much as 59% of subsequent ETH/BTC variation. That level of explanatory power is unusually high, but it is concentrated within a very specific macro and market environment. Removing calendar year 2021 causes the two previously significant regressions to produce p-values of 0.76 and 0.48. Expanding the exclusion window from November 2020 to December 2021 removes any statistically detectable association from those two specifications. The positive full-sample results are therefore heavily influenced by the 2020–2021 period and are not reproduced across the rest of the sample.

(Source: TradingView)

What the dollar changes

Splitting the sample according to the 26-week direction of the US Dollar Index produces one of the clearest contrasts in the analysis.

When the dollar is falling, beta coefficients are positive and p-values are low. In the specification using 26-week windows for both ratios, a 1% increase in copper/gold is associated with an approximately 1.28% increase in ETH/BTC over the following 26 weeks. The p-value is below 0.001.

When the dollar is rising, the coefficient for the same specification falls to −0.31. Its p-value of 0.449 does not support a statistically reliable negative relationship. The more relevant observation is that the positive association seen under a weaker dollar disappears once the dollar strengthens.

This result has a plausible economic interpretation. A weaker dollar can coincide with easier financial conditions, stronger risk appetite and greater demand for cyclical assets. Copper/gold and ETH/BTC may then respond to the same supportive macro environment.

The general dollar control formed part of the original methodology. The precise division between rising-dollar and falling-dollar periods, however, was selected after examining the data. This regime analysis should therefore be treated as exploratory rather than as independent confirmation.

Where the ratios stand now

Current conditions also differ from the weaker-dollar environment in which the relationship appeared strongest. As of 24 Jul, copper/gold had risen by approximately 31.2% over 26 weeks, while ETH/BTC was trading at 0.029. Copper/gold had not yet confirmed a long-term breakout, remaining around 11% below its highest level of the previous two years and approximately 35% below its highest level of the previous three years.

Over the same 26-week period, the US Dollar Index had risen by approximately 4.1%. ETH/BTC was still below 0.03, while bitcoin had fallen by approximately 17.3% over the preceding 13 weeks. The current configuration therefore combines a sharp copper/gold rebound with a strengthening dollar, rather than the weaker-dollar regime associated with the strongest historical results.

Taken together, the evidence points to a conditional macro relationship rather than a stable timing signal. The primary test finds no positive cumulative effect between 13 and 26 weeks. Some exploratory regressions identify a positive association using 26-week copper/gold momentum, but the result depends heavily on the specification and the period examined.

After adjustment for multiple testing, the association no longer meets the 5% significance threshold. It is also heavily concentrated around the 2020–2021 episode and does not translate into reliable out-of-sample forecasts.

Copper/gold can therefore be considered a macro context indicator for ETH/BTC, but not a standalone timing signal.