"Digital gold" is two claims — tested separately
Gold here is PAXG, a token redeemable for allocated physical gold, priced on the same exchange and the same daily UTC candles as the crypto side. That matters for the comparison: a spot gold feed would close on weekends while crypto keeps trading, and the mismatched days would distort a daily correlation badly. PAXG trades continuously, so every day is a like-for-like pair.
The trade-off is that PAXG is a claim on gold rather than gold, and it carries issuer and redemption risk that bullion does not. It tracks the exchange's own gold futures closely — daily returns correlate at roughly 0.99 over their shared period — but it can trade at a small premium or discount, and it is not a substitute for the metal in any argument about counterparty risk.
Volatility is annualised from daily log returns at 365 days per year, matching the rest of this site. Correlations use log returns; the stress rows use simple percentage returns so the averages read directly. The window is whatever the two series share, so it begins when the later of the two started trading.
"Digital gold" is used to mean two things that are rarely separated. The first is that Bitcoin stores value the way gold does — that it holds its purchasing power without much drama. The second is that it protects a portfolio when other things go wrong. These are independent claims, they are tested differently, and an asset can fail one while passing the other.
The first is the easier of the two, and it is not close. Volatility and drawdown are direct measurements requiring no assumptions, and Bitcoin comes out several times more volatile than gold with a far deeper worst loss. That does not make it a bad asset — it has also returned dramatically more over the same period, which is the actual reason most people own it. It just means the stability half of the phrase is describing something Bitcoin has never done.
The second claim needs more care, because a low correlation is often quoted as if it settled the matter. It does not. What matters is the behaviour during stress, not the average across all conditions, and a full-sample correlation blends the two together. The rolling window on this page exists for the same reason: a relationship that swings between clearly negative and clearly positive is not something an allocation can lean on, no matter what its long-run average happens to be.
⚠️ Not investment advice. Gold is represented by a tokenised claim traded on the same exchange, which is not identical to holding bullion. All figures are historical and measured over whatever window the two series share. All decisions and risks are your own.
The phrase bundles two separate claims. On stability it fails clearly — Bitcoin has been several times more volatile than gold with a much deeper worst drawdown. On protection it fails in a subtler way: the correlation is low on average, but gold did not offset Bitcoin’s worst days, which is when a hedge would need to work.
The page measures both from the same daily candles over their shared window. Bitcoin has run around three times gold’s annualised volatility, and its worst peak-to-trough loss has been several times deeper. Those are direct measurements that need no modelling assumptions.
Weakly and positively, not independently. Across the full window the daily correlation is small but its confidence interval excludes zero, so it is a real link rather than no relationship — which is the opposite of what a diversification argument needs.
Because the single number hides how much it moves. Measured over rolling 90-day windows the correlation swings between clearly negative and clearly positive, so whichever value supports an argument, some stretch of history produced it. A relationship that unstable cannot carry an allocation decision.
Not reliably. The page checks gold’s worst days directly and reports what Bitcoin did on each of them — both the average return and how often it rose. That test is the point, because average-condition correlation says little about behaviour during stress.
Because it trades continuously on the same exchange and the same UTC daily candles as crypto. A traditional gold feed closes on weekends while crypto keeps trading, and the mismatched days would distort a daily correlation badly.
No. It is a token redeemable for allocated physical gold, so it carries issuer and redemption risk that bullion does not, and it can trade at a small premium or discount. It tracks the exchange’s gold futures at roughly 0.99 daily correlation, which makes it a sound price proxy but not a substitute in any argument about counterparty risk.