A dollar rally and a Bitcoin sell-off can look like a dependable trading rule. The useful question is whether their relationship survives a different period, measurement method and market backdrop.
Bitcoin and the US dollar do not have a permanent inverse relationship. They may move in opposite directions during one week and rise together during another. Before explaining the pattern, identify the dollar measure, examine shared risk conditions and check for crypto-specific events. Our assessment focuses on evidence that can challenge the explanation, rather than a fixed price target.
Which dollar measure are we comparing?
DXY tracks the dollar against a selected currency basket. EUR/USD is a single exchange rate. Bitcoin quoted in euros reflects both its dollar price and the euro’s value against the dollar. Those are separate effects, so investors looking at different quote currencies can see different returns.
Consider a hypothetical Bitcoin price falling from $60,000 to $57,000 while EUR/USD falls from 1.10 to 1.00. Its euro value rises from about €54,545 to €57,000, a gain of 4.5%, despite the 5% dollar decline. The calculation excludes spreads and fees and uses no current market quote. It illustrates currency translation, not a forecast or evidence about DXY itself.
The Federal Reserve’s broad dollar index must not silently replace DXY in a calculation. FRED’s DTWEXBGS series has a different scope and base. Calling a coefficient calculated from that series a “DXY correlation” misidentifies the result.
What do dated studies actually show?
Table 3 of an August 2023 IMF working paper uses a January 2018–March 2023 sample and reports Bitcoin–dollar index correlations of −0.05 before 2020 and −0.14 afterward. Appendix Table A.2 gives the difference a p-value of 0.060. We do not treat that as a confirmed change at the 5% threshold, or as today’s coefficient.
A negative sign does not mean a strong relationship. −1 represents a perfect inverse linear relationship; a coefficient near zero indicates weak linear co-movement. A correlation coefficient is not a probability: −0.14 does not mean a 14% chance that Bitcoin will fall.
A New York Fed research post dated February 8, 2023 examines short announcement-window responses using a 2017–2022 Bitcoin sample. The authors find no clear systematic response and flag the short sample. An immediate reaction to news and co-movement across months are different questions; one finding does not automatically invalidate the other.
Conditions that can reinforce or break the pattern
A stronger dollar can accompany tighter financing conditions or reduced appetite for risk. Bitcoin selling may be another consequence of the same environment. Instead of assuming that the DXY move directly caused the selling, investigate the shared development. The monetary-policy channel is covered separately in our Fed and crypto analysis.
| Observed setting | Possible explanation | Evidence to check |
|---|---|---|
| Dollar up, BTC down | Reduced risk exposure may appear in opposite directions across the two markets. | Do equities, real yields and funding conditions support that account? |
| Dollar and BTC both up | Crypto-specific demand may outweigh the stronger-dollar backdrop. | Are there verified product flows and spot demand for the period? |
| Dollar and BTC both down | Crypto-specific selling or a confidence shock may occur while the dollar weakens. | What do the event timing, breadth of selling and other crypto assets show? |
| DXY moves, BTC stays flat | The move may be concentrated in particular basket currencies. | Are index components moving, or is the broader risk environment changing? |
These are competing explanations to investigate, not established causes. A single day selected after the event can be made to fit a preferred story. Choose the observation period and comparison measures before seeing the result.
Five checks for a correlation calculation
- Identify the series. Record the BTC/USD source, dollar index name, observation frequency and closing time. If using BTC/USDT, account for the stablecoin’s dollar-price basis.
- Compare returns rather than price levels. A simple return is the latest value divided by the previous value, minus one. Long-running trends can produce misleading similarity between levels.
- Align intervals. Bitcoin trades over weekends. Do not pair a Friday-to-Monday index change with only Sunday-to-Monday Bitcoin performance. Use matching start and end times for both series; do not replace missing observations with zero returns.
- Test window sensitivity. Compare, for example, 30 and 90 shared observations. If the sign changes, do not promote one coefficient into a permanent relationship. Thirty observations are not a universal threshold for adequate evidence.
- Inspect outliers and a later period. Is one large move driving the result? Does the same method hold up on a separate sample? Do not remove an outlier because it spoils the conclusion; report the full and alternative calculations together.
A reproducible note needs the sources, date range, shared observation count, timestamps, return convention and window. A screenshot saying only “correlation −0.70” leaves too much unknown. Overlapping rolling windows are not independent experiments either.
Why inverse correlation does not guarantee a hedge
Direction and magnitude are different questions. Suppose the dollar index rises 1% while Bitcoin falls 8%. In a theoretical example with $1,000 of starting exposure on each side and perfect index tracking, the dollar-side gain is $10 while the Bitcoin loss is $80: a net change of −$70. Real product costs, margin and leverage are excluded.
This is not a proposed hedge. It shows why opposing directions alone do not ensure offsetting gains and losses. Volatility, position sizes and changing relationships matter. The correlation coefficient also does not tell you how many percentage points Bitcoin should move for a 1% index change.
What would change our interpretation?
We would give DXY less explanatory weight if the inverse pattern disappeared across windows, depended on one stress day or was better explained by a crypto-specific event. A pattern that survives different sources and periods would justify examining the shared macro backdrop more closely. It would still not establish causation or a next-day signal.
When a claim relies on an economic release, separate the forecast from the result using the economic calendar guide. A useful assessment identifies the development, interval and evidence, rather than relying on whether the index happens to be green or red.
Check developments in Bitcoin news over the same interval: an ETF decision or network event may need an explanation separate from dollar movements.
If the measure is unfamiliar, start with the DXY definition. The dollar index reading guide explains points, percentages and data types through worked examples.
