Bitcoin Is Down 1.8% in 2026 and Up 2.1% for Canadians
Bitcoin is down 1.81% in US dollars this year. In Canadian dollars it is up 2.08%. Same asset, same calendar, two answers 3.89 percentage points apart, and the entire difference is the loonie.
The US dollar index has just closed at its highest level since April 2025, and the usual reading is that a strong dollar is bad for Bitcoin. For a Canadian holder the dollar’s climb has paid rather than cost. And the record after past dollar highs is too thin to act on in either direction: nine of the twelve episodes since 2016 fall inside the 2018 to 2022 stretch, which goes a long way towards explaining the little apparent weakness the record shows.
As of 7:05 a.m. ET on Monday October 5 2026, Bitcoin (BTC) was at $85,925.70 USD, down 0.64% over 24 hours, and $122,604.81 CAD, down 0.54%. Ethereum (ETH) was at $2,711.69 USD and $3,868.52 CAD, down 0.54% and 0.46%. Live readings rather than closes. Source: Yahoo Finance.
The dollar made a high, and a Canadian owns part of it
The index closed at 102.10 on Thursday October 1 2026, its highest close since April 9 2025, when it finished at 102.90. Friday’s close was 101.93, and Monday morning it was reading 102.218, a live bar rather than a close. The chart is on TradingView’s dollar index symbol page. EUR/USD closed at 1.12499 on October 1, its lowest since May 28 2025 (1.12337), so mechanically this is mostly a euro story.
The index is six currencies, though, and one is ours. ICE’s own U.S. Dollar Index contract FAQ puts the euro’s weight at 0.576 and the Canadian dollar’s at 0.091, fourth behind the yen at 0.136 and the pound at 0.119, and calls the index “a geometrically-averaged calculation of six currencies weighted against the U.S. dollar”.
Those weights are exponents in a geometric mean rather than shares of a move, so we measured the shares. From the end of 2025 to Thursday’s close, on log changes, which is exact for an index built that way, the euro delivered 65.4% of the rise, the Canadian dollar 9.0%, and the other four currencies 25.7% between them. The loonie landing on its weight is a coincidence of this window: it fell by close to as much as the index rose. So this time a dollar-index high was partly a report of a Canadian’s own currency falling.
The same Bitcoin, two returns, 3.89 points apart
| Dec 31 2025 | Oct 5 2026, 7:05 a.m. ET | Change | |
|---|---|---|---|
| Bitcoin in US dollars | $87,508.83 USD | $85,925.70 USD | -1.81% |
| Bitcoin in Canadian dollars | $120,107.79 CAD | $122,604.81 CAD | +2.08% |
| Canadian dollar | 0.730209 USD | 0.701710 USD | -3.90% |
| US dollar index | 98.280 | 102.218 | +4.01% |
| Ethereum in US dollars | $2,967.04 USD | $2,711.69 USD | -8.61% |
Yahoo Finance daily closing prices for BTC-USD and BTC-CAD and our arithmetic on them. The final point on each line is the 7:05 a.m. ET snapshot of October 5 2026, a live reading rather than a close.
The arithmetic is worth showing once. A US-dollar return of -1.81%, converted at a Canadian dollar that fell 3.90%, works out to +2.18%, against the +2.08% that Yahoo’s own BTC-CAD series printed. The 0.10-point difference is the two price feeds disagreeing slightly, not a second mechanism. At either figure the currency leg is the story.
A currency leg that large is either a decision or an accident, and turning it off has a published price. On US equities we measured 90 Canadian depositary receipt series, where the hedge ran a median 2.12% a year in a range from 1.44% to 3.08%. The detail is in our guide to Canadian Depositary Receipts, and what the currency hedge costs.
It is not presentational at tax time either. A gain is assessed in Canadian dollars, so the currency leg sits inside the taxable amount rather than beside it, and half of a capital gain is taxable: see how a capital gain is taxed in Canada.
Two share classes, one Bitcoin portfolio, and what the hedge cost
On Bitcoin itself the same decision sits on the TSX, priced in two traded shares. Purpose Investments lists two exchange-traded classes of one portfolio: BTCC is hedged back into Canadian dollars, BTCC.B is not. Both carry a 1.00% management fee and a 1.29% management expense ratio, and the issuer’s own hedged flag reads Yes against No. Total cost of ownership is identical, so whatever separates their returns, none of it is a fee difference.
| Dec 31 2025 close to Oct 2 2026 close | |
|---|---|
| BTCC, Canadian dollar hedged | -6.74% |
| BTCC.B, not hedged | -0.18% |
| Spread, unhedged minus hedged | +6.56 points |
Dividend-adjusted TSX closes across the 191 sessions in that window. Over the same stretch spot Bitcoin was -3.44% in US dollars and +0.23% in Canadian, a gap of 3.67 points.
The class spread is wider than the spot gap, and the difference is what the hedge cost. The hedged class lagged spot Bitcoin in US dollars by 3.30 points and the unhedged class lagged spot Bitcoin in Canadian dollars by 0.41 points. The same 1.29% MER sits inside both of those lags, so it cancels when one is subtracted from the other: the 2.89 points between them is hedge-related cost over the window. Both lags are exchange prices measured against a spot reference, so a few tenths of a point of slack sits on each, and it is the 2.89 that carries it rather than the 6.56. At that rate the hedge runs roughly 3.8% a year, against the 2.12% median on the equity receipts above, and the slack is wide enough to make that a comparison of orders rather than of decimals.
Neither class is the better one. The hedged class removes a risk that happened to pay this year and would have cost in a year the loonie rose, and the measured price of removing it over this window was 2.89 points.
What a dollar high has actually meant for Bitcoin
Start with the fact that cuts against the usual reading. From the September 2 2026 close, the stretch over which the dollar actually rallied, the index rose 2.67% and Bitcoin rose 11.16% in US dollars, 14.54% in Canadian with the loonie down 2.85%. Ethereum rose 13.39% in US dollars. The dollar went up and Bitcoin went up with it. One month is not a record, so here is the record.
Bitcoin’s daily closes, aligned to the dollar index’s trading days, give 3,031 sessions from September 17 2014 to October 5 2026. An episode is a day the index closes at its highest level of the trailing 252 sessions, counted once per 60 sessions so one long dollar run is not read as a dozen signals. That gives 13 episodes, and the thirteenth is October 1 2026, with no forward history yet. Twenty sessions is roughly a trading month, sixty roughly three, and each episode is measured in both currencies.
| Episode | 20 sessions, USD | 20 sessions, CAD | 60 sessions, USD | 60 sessions, CAD |
|---|---|---|---|---|
| Nov 15, 2016 | +9.82% | +8.44% | +39.21% | +35.28% |
| Aug 9, 2018 | -1.54% | -0.73% | -2.74% | -2.17% |
| Nov 12, 2018 | -45.27% | -44.81% | -42.74% | -42.46% |
| Mar 7, 2019 | +25.86% | +25.12% | +109.87% | +109.83% |
| Jul 31, 2019 | -3.28% | -2.47% | -25.70% | -26.40% |
| Feb 18, 2020 | -48.48% | -44.66% | -8.60% | -2.83% |
| Sep 29, 2021 | +40.70% | +36.47% | +22.18% | +22.88% |
| Jan 27, 2022 | +5.59% | +5.31% | +8.94% | +8.86% |
| Apr 25, 2022 | -28.08% | -27.80% | -42.74% | -42.11% |
| Aug 22, 2022 | -11.72% | -9.58% | -21.10% | -19.73% |
| Nov 13, 2024 | +10.44% | +12.20% | +5.70% | +7.92% |
| Jun 18, 2026 | +3.71% | +3.20% | +20.22% | +18.38% |
ICE U.S. Dollar Index (DXY) and BTC-USD daily closes from Yahoo Finance and our arithmetic on them. An episode is a DXY close at the highest level of the trailing 252 sessions, counted once per 60 sessions.
Six up and six down, at both horizons. Against how Bitcoin behaved on any day in the same history, the conditional record is mildly worse and nothing more.
| After a dollar-index one-year high | Bitcoin on any day | |
|---|---|---|
| Median, next 20 sessions, US dollars | +1.09% | +2.92% |
| Median, next 60 sessions, US dollars | +1.48% | +7.87% |
| Median, next 20 sessions, Canadian dollars | +1.24% | +2.79% |
| Median, next 60 sessions, Canadian dollars | +2.88% | +7.07% |
| Positive, next 20 sessions, US dollars | 6 of 12 | 57.3% |
| Positive, next 60 sessions, US dollars | 6 of 12 | 58.0% |
The means point the other way and are the less useful number: -3.52% at twenty sessions, dragged below zero by the two episodes that lost more than 45% in a month, and +5.21% at sixty, lifted by March 2019’s +109.87%. With twelve observations one episode moves the average, which is why the median leads. The worst three-month outcome was -42.74%, recorded twice, after November 12 2018 and April 25 2022, and the best was that +109.87%.
The Canadian cushion after a dollar high is real and small
A dollar high usually coincides with a weaker loonie, so it is tempting to assume a Canadian is systematically cushioned after one. Measured, the cushion exists and barely registers. Episode by episode the Canadian return beat the US return by a median of 0.37 points at twenty sessions and 0.43 points at sixty, and was the better of the two in 7 of 12 episodes at both horizons. After the other five the loonie rose and the Canadian holder did worse rather than better. February 2020 was the largest cushion at both horizons, worth +3.82 points at twenty sessions and +5.77 at sixty, when Bitcoin fell 48.48% in US dollars and 44.66% in Canadian. The largest drag was September 2021 at twenty sessions, at -4.23 points, and November 2016 at sixty, at -3.93 points.
Over the year the currency moved 3.89 points. Over the month after a dollar high it moved 0.37, and after five of the twelve it moved against the holder.
Three thresholds, two null models and twenty-four tests
One threshold is not a test of anything, so we ran three: a six-month high, the one-year high above, and an eighteen-month high. The null model needed widening too. Nine of the twelve episodes fall between 2018 and 2022, with only November 2016, November 2024 and June 2026 outside that stretch, so a permutation test scattering placebo dates uniformly across twelve years is partly asking how Bitcoin did in those five years against the whole period, then crediting the answer to the dollar. The second null holds the era constant: each placebo date is drawn from the same calendar year as the episode it replaces, which asks whether the dollar high adds anything on top of the year it fell in.
| Definition of a dollar-index high | Episodes | Horizon | Episode median | Positive | p, uniform null | p, era-matched null |
|---|---|---|---|---|---|---|
| Six-month high (126 sessions) | 18 | 20 sessions | +2.35% | 10 of 17 | 0.45 / 0.64 | 0.60 / 0.75 |
| Six-month high (126 sessions) | 18 | 60 sessions | -8.60% | 6 of 17 | 0.07 / 0.05 | 0.20 / 0.15 |
| One-year high (252 sessions) | 13 | 20 sessions | +1.09% | 6 of 12 | 0.37 / 0.41 | 0.51 / 0.53 |
| One-year high (252 sessions) | 13 | 60 sessions | +1.48% | 6 of 12 | 0.32 / 0.39 | 0.64 / 0.71 |
| Eighteen-month high (378 sessions) | 8 | 20 sessions | -3.82% | 3 of 8 | 0.15 / 0.22 | 0.14 / 0.19 |
| Eighteen-month high (378 sessions) | 8 | 60 sessions | -5.75% | 3 of 8 | 0.17 / 0.21 | 0.20 / 0.23 |
Each p-value cell reads as the probability of a median this low, then of this few positive. The six-month and one-year thresholds each score one fewer episode than they count, because the most recent one has no forward history yet. October 1 2026 is not an eighteen-month high, so that row scores all eight. The p-values are Monte Carlo estimates from 20,000 draws, so two decimals is as far as the arithmetic goes. In plain English, for the row this piece is built on: pull twelve dates out of a hat instead of twelve dollar-index highs and you would get a result this weak about a third of the time.
Then the result that matters, and its direction is the surprise. At sixty sessions the uniform placebo median is about +8%, which is not the +7.87% in the table above: one is the median of every sixty-session return in the history, the other the median of the medians of twelve-date draws. Hold the era constant and the placebo median drops to about -2%, below the episodes’ own +1.48%. Both are resampling estimates quoted to the precision they carry: across five independent seeds at 40,000 draws the uniform figure moved by a tenth of a point and the era-matched one by a little over two tenths, so the sign flip and the ordering survive anywhere in those ranges. What that licenses is narrow and still worth having: the high, as a dated event, adds nothing on top of the year it fell in.
Twenty-four tests in all. The smallest p-value is 0.05, on the six-month threshold sixty sessions forward under the uniform null, and it weakens to 0.15 once the era is held constant. Every other test sits at 0.07 or above, and 22 of the 24 at 0.14 or above. One test in twenty-four landing at 0.05 is about what chance produces when there is nothing to find, so that cell is not evidence of an effect either, and it weakens further, like every other cell, once the era is held constant.
So the conclusion is not that a strong dollar is good for Bitcoin, and not that it is bad. The record is too thin to separate either from noise.
The correlation cuts the other way
One finding genuinely pushes back. The daily-return correlation between the dollar index and Bitcoin is -0.073 across the whole 3,031-session history, barely anything, and -0.151 over the last 252 sessions. Over the last 90 sessions it is -0.354, the 6.3rd percentile of all 2,941 rolling 90-day windows since 2014, so the inverse relationship is unusually tight right now. Those windows run from -0.512 (ending January 12 2023) to +0.274 (ending January 23 2018), only 65.5% are negative at all, and the median is -0.062.
So the daily relationship is unusually tight while the episode record says the high carries no forward information. Both are true, and they answer different questions. A negative daily correlation says the two series’ day-to-day moves lean in opposite directions. A negative trend would say one ends up lower when the other ends up higher. Across the last 90 sessions those daily moves have leaned against each other more strongly than usual, and over the September 2 stretch inside that window both still finished higher.
One reading fits the record without being measured by it. The two worst outcomes, November 2018 and April 2022, plus February 2020, sit inside bear markets or the pandemic crash, and the two best, March 2019 and September 2021, sit inside bull markets. That is consistent with the index and Bitcoin both reading the same risk appetite rather than either driving the other, and the era-matched null points the same way. It is a hypothesis the record does not contradict, not proof of it.
The honest limits of that record
Twelve episodes is a small sample, and nothing above should be read as if it were a hundred. The history starts in 2014 and spans two halvings, an ETF launch and a near-total change in who owns the asset, so the 2016 and 2018 episodes come from a different asset in all but name. The 252-session rule is one of several reasonable definitions of a one-year high, which is why there are three in the grid. The era-matched null is likewise one choice among several, calendar year is a coarse proxy for a market regime, twelve episodes spread across twelve years leaves very little inside any one of them, and for the 2026 episode the matched year is really its first half, since at sixty sessions the last date with a forward return falls in early July. And the October 1 2026 episode is live and unresolved, which is the one every reader of this is inside.
What a Canadian holder can take from it
The dollar index high is not an instruction in either direction. Across three thresholds and two null models it tells you close to nothing about Bitcoin’s next month or quarter, and the month just past, with the index up 2.67% and Bitcoin up 11.16% in US dollars, showed it.
The currency is the part that does real work, and it works whether or not anybody chose it. A 3.89-point gap between two measurements of the same asset over nine months is not a rounding artifact, and over a long holding period that term compounds alongside the asset. The 6.56 points between BTCC and BTCC.B is the size of that decision rather than its price: in a year the loonie fell the unhedged class collected it, and in a year the loonie rose it would have paid it away. Turning the exposure off cost 2.89 points. The question lands differently again on the equities, where some Canadian crypto names trade in Canadian dollars on the TSX and TSXV and others in US dollars on the Nasdaq, which changes a holder’s exposure before the business is even considered. That is the lens behind our list of Canadian crypto stocks ranked on Bitcoin per share.
Disclaimer: The content on bestcanadianstocks.ca is for informational and entertainment purposes only and does not constitute financial advice. Past performance is not indicative of future results. Always consult a qualified financial advisor before making investment decisions. Bitcoin and Ethereum prices are a Yahoo Finance snapshot taken at 7:05 a.m. ET on October 5, 2026. Crypto trades continuously, so the daily bar was still open and the snapshot is a live reading rather than a close. The US dollar index is the ICE U.S. Dollar Index on Yahoo Finance, daily closes except for the Monday reading of 102.218, which is a live bar pulled at 7:08 a.m. ET. The index weights and the description of the index as a geometric average are from ICE Futures U.S., U.S. Dollar Index Contracts FAQ, page 2. The contribution shares are the exact log-change decomposition of that geometric index from the December 31, 2025 close to the October 1, 2026 close. Returns, the episode study, the medians, the hit rates and the correlations are our own arithmetic on Yahoo daily closes for DX-Y.NYB, BTC-USD, BTC-CAD, CADUSD and ETH-USD, 3,031 sessions from September 17, 2014 to October 5, 2026, with Bitcoin aligned to the dollar index’s trading days. An episode is a dollar index close at the highest level of the trailing 252, 126 or 378 sessions as labelled, counted once per 60 sessions. Both permutation tests use 20,000 draws on seed 20261005: the uniform null draws placebo dates from the whole history, and the era-matched null draws each placebo date from the same calendar year as the episode it replaces. P-values are Monte Carlo estimates and are reported to two decimals. The BTCC and BTCC.B figures are dividend-adjusted TSX closes over the 191 sessions from the December 31, 2025 close to the October 2, 2026 close, a window that ends on Friday rather than Monday because both are exchange-listed. Their identical 1.00% management fee and 1.29% MER, and the hedged flag that reads Yes on BTCC and No on BTCC.B, are from Purpose Investments’ own Purpose Bitcoin ETF fund page, fetched October 5, 2026. The Canadian depositary receipt hedge costs are from our own published guide. The conversion of Bitcoin’s US dollar return at the Canadian dollar’s move is arithmetic on the two rates and lands 0.10 points from the BTC-CAD series on the same feed.
