Gold had its worst week in months: COMEX futures fell 4.1%, from $4,181.10 on 16 October to $4,007.90 on 23 October (Yahoo Finance), with the heaviest losses on Thursday (−2.2%) and Friday (−1.3%). Bitcoin, which supporters have called “digital gold” for a decade, lost only 0.5% Friday to Friday over the same dates. That looks like proof of independence, but the daily record is less flattering to the idea. In the past year, on the days gold fell 1% or more, bitcoin averaged −1.39%; on the days gold rose 1% or more, it averaged +0.87%. A gold hedge would show the opposite pattern.
The test below uses daily closes for gold futures (GC=F) and bitcoin (BTC-USD) from Yahoo Finance and compares each bitcoin return with the gold return over the same gold trading days. We do not claim causation, and the main finding is a modest same-direction tendency, not a rule.
The numbers
| Sample / condition | Days | Bitcoin average | Bitcoin median | Days bitcoin rose | Standard error |
|---|---|---|---|---|---|
| Past year, all days | 251 | −0.08% | −0.19% | 117 | 0.17 pts |
| Past year, gold ≤ −1% | 59 | −1.39% | −1.60% | 15 | 0.36 pts |
| Past year, gold ≥ +1% | 59 | +0.87% | +0.70% | 38 | 0.40 pts |
| Past year, gold within ±1% | 133 | +0.07% | — | — | 0.20 pts |
| Five years, all days | 1,257 | +0.07% | −0.05% | 618 | 0.09 pts |
| Five years, gold ≤ −1% | 185 | −0.60% | −0.51% | 72 | 0.28 pts |
| Five years, gold ≥ +1% | 217 | +0.63% | +0.49% | 131 | 0.24 pts |
Three things follow. First, the direction is the same in both samples: bitcoin has fallen more often than it rose when gold had a big down day (15 of 59 days up in the past year, 72 of 185 over five years) and risen more often when gold had a big up day (38 of 59 and 131 of 217). Second, the effect is much stronger in the last year than over five years, a gap of 0.8 percentage points on the down-days that is larger than either standard error, which tells us the relationship is not stable. Third, the overall daily correlation is small: 0.23 over the past year and 0.11 over five years, which means gold explains very little of bitcoin’s day-to-day variation.
Stricter thresholds
| Past year, gold move | Days | Bitcoin average | Days bitcoin rose |
|---|---|---|---|
| ≤ −2% | 24 | −1.59% | 4 |
| ≤ −1.5% | 31 | −1.81% | 4 |
| ≤ −1% | 59 | −1.39% | 15 |
| ≥ +1% | 59 | +0.87% | 38 |
| ≥ +1.5% | 42 | +0.91% | 27 |
| ≥ +2% | 27 | +0.57% | 16 |
The 4-of-24 and 4-of-31 counts are the most striking: on the biggest gold down-days of the year, bitcoin rose on only four. But 24 and 31 days are small samples, the days overlap with broad risk-off episodes in which most assets fell together, and we picked these cut-offs after looking at the numbers, so the pattern should be read as descriptive. The five-year version of the 1.5% test (92 days) shows a milder −0.47% average, with 35 days up.
Does the pattern change by year?
Yes. Counting only gold down-days of 1% or more, bitcoin’s average was −0.96% in 2022 (37 days), +0.95% in 2023 (24 days), −0.52% in 2024 (31 days), −0.09% in 2025 (35 days) and −1.50% in 2026 so far (53 days). Only 2023 showed a clearly positive average on gold’s bad days, which would be the hedge profile (2021’s five qualifying days averaged +0.01%). The 2026 reading is the weakest for the digital-gold story and coincides with a period when bitcoin has traded more like a high-beta stock; see our 30-day Nasdaq correlation study, where the reading was 0.60.
What about the whole week?
On a weekly basis the picture is murkier. Across 252 Friday-to-Friday weeks since October 2021, we found 22 weeks in which gold fell 3% or more. Bitcoin averaged −0.56% in those weeks, with 10 of the 22 up, against an average of +0.36% across all weeks. Recent examples show the range: the week to 5 June 2026 saw gold −5.0% and bitcoin −17.0%; the week to 20 March saw gold −9.6% and bitcoin −0.6%; the week to 2 October saw gold −3.7% and bitcoin +0.6%; and last week saw gold −4.1% and bitcoin −0.5%. Twenty-two observations cannot support a firm conclusion, and the cases where gold fell hardest include a period (March) when bitcoin barely moved.
Why gold fell, and why that matters for bitcoin
The chain linking gold and bitcoin this week runs through rates. Markets are pricing a possible Fed hike by December (Reuters: roughly 25 basis points before year-end), and the Fed decides on Wednesday 28 October with September PCE inflation due the next day. Higher expected rates tend to hurt non-yielding assets, and both gold and bitcoin pay no interest. That is a plausible shared driver, not a demonstrated one: we did not test it, and gold’s fall coincided with a surge in oil to $116.72 Brent and a sharp equity selloff, both of which have their own effects. Our earlier results on the oil link and the Fed-day record give more context, and our guide to the PCE report explains what to look for on Thursday.
How to use this
- Do not treat bitcoin as a hedge against a gold selloff. In the last year it has moved with gold on large days far more often than against it, though the effect is modest.
- Check the regime. The same test run on 2023 would have supported the digital-gold story; run on 2026 it does not. Re-run it quarterly.
- Mind the clocks. Gold futures and bitcoin close at different times, so a few hours of each day do not overlap.
Method and limits. Daily closes for gold futures (GC=F) and BTC-USD come from Yahoo Finance. For each gold trading date from 26 October 2021 to 23 October 2026 (1,257 dates with data in both series) we computed gold’s return from the previous gold trading date and bitcoin’s return between the same two dates, so a Monday observation covers the weekend and a post-holiday observation spans several days. The “past year” covers 251 observations after 23 October 2025. Standard errors are the sample standard deviation divided by the square root of the count. The ±1% and ±1.5% thresholds were chosen after looking at the data, and the yearly and weekly splits are exploratory. Weekly figures are Friday closes from the same series. Gold futures settle during the US afternoon; bitcoin’s Yahoo close is at 00:00 UTC. Yahoo’s futures series is a continuous contract and may include roll effects.
Method: bitcoin prices, highs, lows and closes are Coinbase Exchange candles (UTC day) pulled by Bitcoin Mastery at about 20:30 UTC on Sunday 25 October 2026; the 25 October candle is still open at that time, so any closing figure quoted is the 24 October close unless stated. Index, stock, gold and oil figures are Yahoo Finance daily data (US closes of Friday 23 October) or the news sources cited. Other third-party figures are attributed to their source and date. All averages, medians, counts, correlations and standard errors above are our own calculations.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. Bitcoin and other cryptocurrencies are volatile and you can lose some or all of the money you put in. Nothing here is a recommendation to buy, sell or hold bitcoin, any exchange-traded fund, any listed security or any other asset, and the technical levels, probabilities and scenarios discussed are descriptions of published data, not forecasts. Do your own research and consult a licensed financial advisor before making investment decisions.