Is bitcoin any use for trading forex?
Is there a real correlation between bitcoin and forex? We measured 1,449 days: it predicts nothing at any horizon, but strip the dollar out and +0.187 survives against the Australian dollar — risk appetite, not a currency effect.
Bitcoin trades around the clock, moves before almost everything else, and shares a screen with forex just about everywhere. The question almost asks itself: does it warn you about anything? We measured it across 1,449 days and 31,681 hours of our own bars. It does not warn you — but unlike gold, it is not just the dollar in disguise either.
First, the obvious: it does move with forex
On daily closes from 2021 to 2026, bitcoin returns and EUR/USD returns are correlated at +0.175. Against the Australian dollar, +0.268. Against sterling, +0.219.
Modest numbers next to gold's +0.393, but real. And the explanation is textbook: bitcoin is priced in dollars too, so a softer dollar pushes it up through the same arithmetic that pushes EUR/USD up.
So what is left once the dollar is removed?
This is where bitcoin parts company with gold, and it is the interesting finding in all of this.
Strip the dollar's effect out of both series — that is partial correlation — and measure again:
| Bitcoin against… | Raw | Dollar removed | p |
|---|---|---|---|
| EUR/USD | +0.175 | −0.010 | 0.70 |
| GBP/USD | +0.219 | +0.082 | 0.002 |
| USD/JPY | −0.017 | +0.133 | <0.0001 |
| AUD/USD | +0.268 | +0.187 | <0.0001 |
Against the euro, bitcoin dies exactly as gold does: −0.010, indistinguishable from zero. There was nothing there but the greenback.
But against the Australian dollar and the yen it survives, with room to spare. Look at the USD/JPY row: raw, there was no apparent relationship at all (−0.017), and removing the dollar makes one appear at +0.133. That happens when two opposing effects were cancelling out — the dollar effect was masking the other one.
What is that other thing?
Risk appetite, and the proof sits in the crosses that carry no dollar — which, needing no control at all, cannot be an artefact of the control:
| Bitcoin against… | Correlation |
|---|---|
| EUR/JPY | +0.128 |
| EUR/CHF | +0.105 |
| EUR/GBP | −0.100 |
Bitcoin up coincides with a weak yen and a weak Swiss franc. Those two are the market's funding havens: when risk appetite returns they get sold to finance positions in riskier assets, and bitcoin is among the most volatile things quoted anywhere. The third row says the same from the other side: sterling is more cyclical than the euro, so when bitcoin rises the euro loses ground against it.
That said, "risk appetite" already has a gauge, and the fair question is how much bitcoin adds on top of it. Adding the VIX volatility index and US yields as further controls:
| Bitcoin against… | Dollar removed | …and the VIX | …and yields |
|---|---|---|---|
| AUD/USD | +0.187 | +0.128 | +0.111 |
| USD/JPY | +0.133 | +0.098 | +0.067 |
The relationship holds, but the VIX takes a third of it and yields take a little more. A good part of what we are calling risk appetite is risk appetite that was already in the VIX; what bitcoin contributes on its own is about three fifths of what the headline number suggests.
Then does it lead?
No. That part is identical to gold.
| Bitcoin moves… | Correlation with EUR/USD afterwards |
|---|---|
| 1 hour earlier | −0.014 |
| 2 hours earlier | −0.003 |
| 3 hours earlier | +0.005 |
| 6 hours earlier | +0.009 |
| 12 hours earlier | −0.009 |
| 1 day earlier | −0.043 |
Nor against the Australian dollar, where a contemporaneous relationship genuinely exists: the best lag over the training period was +0.013 at six hours, and in 2025-2026 it came in at −0.002. Carrying the best lag of all twelve combinations we tested — gold, silver and bitcoin against four pairs — into the later period, not one repeated.
The one exception, and it lasts seconds
There is one place where bitcoin does lead, and it is exactly the one logic suggests: it trades on Saturday and Sunday, and forex does not.
We measured bitcoin's move from Friday's close to Sunday — entirely observable before forex reopens — against what the pair does on Monday. Across 288 Mondays:
| Monday in… | Correlation with bitcoin's weekend |
|---|---|
| AUD/USD | +0.154 (p = 0.009) |
| GBP/USD | +0.098 |
| EUR/USD | +0.095 |
| USD/JPY | +0.039 |
The Australian dollar one is real, and it replicates: +0.137 over 2021-2024 and +0.239 over 2025-2026. It is the only predictive result in the entire study.
And it cannot be traded. Split Monday in two — the reopening gap and the session that follows — and the effect sits entirely in the gap:
| AUD/USD, against bitcoin's weekend | Correlation |
|---|---|
| Reopening gap | +0.234 (p = 0.0001) |
| Monday session, market already open | +0.042 (p = 0.48) |
The market reopens with bitcoin's weekend already in the price. By the time you can click there is nothing left: the session gives +0.042, and out of sample it falls to −0.029. Bitcoin does not tell you what Monday will do; it tells you where Monday will open, which is information with a shelf life of zero seconds.
A warning: it is not a constant
The +0.187 is an average over nearly six years, and there is a lot of movement behind it. Year by year, bitcoin's partial against the Australian dollar runs from +0.266 in 2022 to +0.042 in 2023, a full year in which it simply did not exist (p = 0.50). Against the yen, so far in 2026 it stands at +0.024 (p = 0.77), and two of the three falsification crosses have flipped sign this year.
That does not invalidate the average — five of the six years point the same way — but it does constrain what you can do with it: this is a property of the regime, not a constant of bitcoin.
So how do you use it?
As a thermometer, not a signal — and a noisier one than the VIX itself, which in these same data reads −0.308 against the Australian dollar versus bitcoin's +0.187. Bitcoin's advantage is not precision: it is that it trades at the weekend and the volatility index does not.
Two uses that follow from what we measured, and one that does not:
- Hidden-correlation warning. This is the solid one. If you are simultaneously long AUD/USD, short USD/JPY and holding crypto, that is not three ideas — it is one bet on risk appetite, in the same vein as what we covered about trading two correlated pairs.
- Regime confirmation, with expectations adjusted. Bitcoin and AUD/USD share a common factor, but a small one: +0.187 means 3.5% of variance explained and agreement on direction 56% of days. One day of disagreement says nothing — they disagree nearly half the time. Only a large, sustained divergence would be informative, and we have not measured that as a signal.
- What does NOT follow: calling Monday. We just measured it, and the market has already priced it into the opening gap.
The bottom line
Bitcoin is no use for predicting forex. Not at any horizon from one hour to five days, not in any pair we measured, with nothing that repeats outside the period where it was found — and the one thing that does lead, the weekend, is spent in the first instant of the reopen.
It is useful for something else, and that something is not trivial: it is a risk gauge that is always trading. Against the euro it adds nothing the dollar index does not already give you; against the Australian dollar and the yen it is telling you something real — though modest, and not always present — about the market's mood. Which is exactly the opposite of what happens to gold, where removing the dollar leaves absolutely nothing behind.
How this was measured: our own bars from January 2021 to August 2026, sampled at the 17:00 UTC close — 1,449 daily and 31,681 hourly observations. Bitcoin pricing from Bitstamp, which does trade at weekends. "The dollar" is an index we build from the remaining major pairs in each case, not the DXY. The hourly figures in the lag table come from the 2021-2024 estimation window; the daily row and the weekend figures use the full sample. It is a sample, not a law of the universe.
