Research note 01 · Crypto risk · 4 min read
Does Ethereum diversify Bitcoin downside?
The 50/50 mix increased risk. The reason is in the covariance, not the number of assets.
A second asset can add exposure without adding protection. Across 2024–2025, a portfolio rebalanced daily to 50% Bitcoin and 50% Ethereum suffered a 48.1% closing drawdown, versus 32.1% for Bitcoin. Its annualized volatility was also higher: 56.1% versus 48.1%.
The question is not whether the assets move identically. It is whether adding ETH reduces risk relative to holding BTC alone. We test that benchmark using 731 paired daily returns from Coinbase Exchange.
01 / The loss path is the first test
The mix’s largest peak-to-trough loss was 16.0 percentage points deeper. We reconstruct each portfolio’s daily wealth and measure its decline from its own previous closing peak; averaging the assets’ maximum drawdowns would give the wrong comparison.
More assets. A deeper drawdown.
Decline from each portfolio’s previous closing peak · 2024–2025
02 / Higher volatility overwhelmed diversification
ETH’s annualized volatility was 70.2%, against BTC’s 48.1%. Their daily returns had a 0.79 correlation: imperfect co-movement, but substantial shared risk.
For this fixed 50/50 allocation to be less volatile than BTC, correlation would have needed to be below 0.30, holding the observed asset volatilities constant. It was nowhere close. The diversification benefit existed, but was too small to offset ETH’s greater volatility.
Correlation below one is insufficient. The added asset must reduce the risk of the portfolio you actually hold.
03 / Stress days and annual checks agree
On Bitcoin’s 37 worst days—the bottom 5% of this sample—ETH fell on 36. The mix lost 6.0% on average on those days, versus 5.3% for BTC. ETH generally reinforced the loss when protection mattered most.
Splitting the sample by calendar year preserves the 50/50 comparison. The size of the disadvantage changes; its direction does not.
| Annualized volatility | Bitcoin | 50/50 BTC + ETH |
|---|---|---|
| 2024 | 53.4% | 56.2% |
| 2025 | 41.9% | 56.0% |
Conclusion
In this sample, a 50/50 BTC–ETH portfolio increased Bitcoin downside risk. The evidence agrees across the loss path, stressed days and both annual volatility checks. ETH added a different asset, but did not provide effective downside protection against the BTC benchmark.
This is a historical result for this allocation, not a forecast or a claim about every ETH weight. A future diversification case needs evidence of lower portfolio risk—not simply two ticker symbols.
Data, method & reproduction
Sample. 1 January 2024–31 December 2025 UTC. Six saved Coinbase Exchange BTC-USD/ETH-USD responses supply 732 consecutive closes per asset, including the initial 31 December 2023 close. Retrieved 8 September 2026. No missing retained dates or conflicting overlaps.
Calculations. Close-to-close simple returns; sample standard deviation × √365; daily equal-weight rebalancing; wealth starts at one. The correlation ceiling follows from setting ¼σ²BTC + ¼σ²ETH + ½ρσBTCσETH below σ²BTC. Tail days are the lowest ceil(0.05 × 731) BTC returns, selected retrospectively. Annual checks reset the sample.
Limits. One venue, two assets and two years. Closing prices omit intraday losses; the portfolio excludes fees, spreads, slippage and taxes. These checks share the same dataset and are not independent out-of-sample validation. The 36/37 count is not a future probability estimate.
Coinbase candle definitions · Daily closes (CSV) · Results (JSON) · Covariance calculation (JSON)
Download the notebook, raw data and independent audit. The notebook reruns the calculations and exposes calendar-window and cost assumptions. See the included README for instructions.