Bitcoin’s Uptober Failed Twice in a Row. February Beats It on Two Measures of Three.
Bitcoin (BTC) is on course to finish October down 7.28% in US dollars, its second red October in a row and the first back-to-back pair in its record. “Uptober” is crypto’s most-quoted seasonal, and the record behind the name was twelve Octobers with a mean of +16.63% and nine gains in twelve. The interesting part is not that it failed. It is how little it took: one month knocked nearly two points off both the mean and the median, and the hit rate from 75% to 69%.
Every price here comes from one pull of Yahoo Finance daily closes at 8:07 a.m. ET on October 31, 2026. Bitcoin’s month does not end until midnight UTC, so the October 31 figures are a live bar rather than a close, with about twelve hours of trading left to run.
The tape, and what October did
Bitcoin trades at $77,469.21 USD, or $108,260.55 CAD, and Ethereum (ETH) at $2,245.38 USD, or $3,137.92 CAD. Against the September 30 closes that is -7.28% for Bitcoin and -16.33% for Ethereum in US dollars. Bitcoin came into the month off its best quarter since 2024, up 42.68% in the third quarter, so this is a 7% October after a 42% run rather than one in a flat tape.
Bitcoin sits 37.9% below its highest daily close on record, the $124,752.53 USD of October 6, 2025. That high was set inside an October, and October 2025 still finished down 3.95%: from the October 6 close to the October 31 close, Bitcoin fell 12.2% in the month that printed its record.
Most of this month’s damage is recent. The best October close was $86,480.30 USD on October 4, the worst $77,198.21 USD on October 30. Through October 28 the month was down 2.74%, ordinary drift. From that close to this morning’s live bar Bitcoin fell 4.67%, so roughly two-thirds of October’s decline landed in the final three sessions. Ethereum’s shape matches: high close $2,726.51 USD on October 4, low $2,241.05 USD on October 30.
The record that earned the name
The twelve completed Octobers from 2014 through 2025 are what people were quoting when this month began. Mean +16.63%, median +12.94%, nine of twelve positive for a 75% hit rate, best +49.09% in 2017 and worst -12.57% in 2014. Bitcoin averaged +4.74% across all other months, putting October ahead by 11.89 percentage points.
That gap looks like it passes a test, and the test is the wrong one. Shuffling which months get called October 100,000 times over the pre-2026 sample, a one-sided permutation puts the chance of a gap that large arising by luck at p = 0.032. But October was never nominated in advance. It is the best of twelve candidate months, and the question worth asking is whether the best of twelve looks like this by chance. Rerun the same shuffles taking the largest gap any month produces, and the answer is p = 0.345. A gap that size turns up about a third of the time in data with no seasonality in it at all. The confidence interval says the same thing from the other direction: 95% on that October mean runs from +4.43% to +28.84%, so “October averages 17%” and “October averages 5%” both sit inside what twelve observations can establish.
Year by year, the record is two enormous green Octobers, a run of modest ones, and a red pair at the end.
The red ones before this year, 2014, 2018 and 2025, were each isolated. 2025 and 2026 are the first back-to-back pair in the record.
Now fold this month in. Entering October, those twelve Octobers ranked first of twelve by mean, first by median ahead of February’s +11.86%, and tied first on hit rate with February and July. First or joint first on all three. One unfinished month moved October to first, second and third: the mean down 1.84 points to +14.79%, the median down 2.02 points to +10.92%, the hit rate from 75% to 69%, nine of thirteen. The claim in this article’s own headline was not true a month ago. That is what a thirteen-sample base rate is worth.
February beats it, and nobody says Februpump
Here is Bitcoin’s full record by calendar month, October running to this morning’s live bar.
| Month | n | Mean | Median | Positive | Hit rate |
|---|---|---|---|---|---|
| January | 12 | -1.15% | -3.46% | 6 | 50% |
| February | 12 | +10.19% | +11.86% | 9 | 75% |
| March | 12 | +0.48% | -0.16% | 6 | 50% |
| April | 12 | +10.16% | +9.70% | 8 | 67% |
| May | 12 | +8.08% | +3.37% | 6 | 50% |
| June | 12 | +0.05% | -0.51% | 6 | 50% |
| July | 12 | +8.88% | +8.13% | 9 | 75% |
| August | 12 | +1.95% | -7.19% | 4 | 33% |
| September | 12 | -1.14% | -0.24% | 6 | 50% |
| October | 13 | +14.79% | +10.92% | 9 | 69% |
| November | 12 | +7.51% | +7.58% | 7 | 58% |
| December | 12 | +7.14% | -3.16% | 5 | 42% |
October ranks first by mean. It ranks second by median, behind February. It ranks third by hit rate, behind February and July, which both sit at 75%. February beats it on two of the three measures and has no nickname.
February is not vindicated by any of this either. October’s gap over the other months is the largest in the table, so once it fails the best-of-twelve test, nothing smaller clears it.
The measure October does win is the fragile one. Its mean is carried by two years: take out 2017 at +49.09% and 2021 at +40.03% and it falls to +9.38%, below February’s +10.19%. A mean over thirteen observations with a standard deviation of 19.6 points is mostly a report on its two biggest years.
What separates October from February is that one of them got a name. A month boundary is a convention, a line drawn in a calendar, not a mechanism in the market. The same is true of correction and bear market thresholds, round numbers that describe how far something fell and say nothing about why, or about what comes next.
Ethereum never had much of an October
If “Uptober” described something about crypto rather than something about Bitcoin, the second-largest asset in the market should show the same shape. By median it does not: October is Bitcoin’s second month and Ethereum’s fifth.
Ethereum’s nine Octobers average +4.17% with a median of +2.28%, five of nine positive. Its 98 non-October months average +4.97%, so on the mean there is no October premium at all. On the median the premium is thin rather than absent, +2.28% against +0.63% elsewhere, enough to put October fifth of twelve and in the upper half, but well behind July at +11.50%. Ethereum is now on its third consecutive red October after 2024 and 2025. Nine observations is a thinner sample than the thirteen it is being used to question, which is the honest limit of this check.
It is still the check that matters most, because the two assets move together almost all of the time. A seasonal carrying something real about crypto would show up clearly in both. One that is an accident of Bitcoin’s particular thirteen years shows up clearly in one.
Bitcoin tracked the Nasdaq. The TSX went the other way.
Every series below runs September 30 close to October 30 close, the last session all of them share, from our own Yahoo pull:
| Series | October 2026 |
|---|---|
| S&P/TSX Composite | +1.09% |
| S&P 500 | -0.93% |
| Nasdaq Composite | -5.30% |
| Bitcoin, USD | -7.61% |
| Ethereum, USD | -16.49% |
Bitcoin trailed the S&P/TSX Composite by 8.70 percentage points in the month that is supposed to be its best. Against the Nasdaq Composite the gap is 2.31 points, which is close enough to call it the same month: Bitcoin roughly tracked the US tech tape, and the series going its own way in that table is the TSX. Ethereum is the real outlier, 11.19 points behind the Nasdaq. For readers who hold the theme through the stock market instead of the coins, our ranked Canadian crypto stocks page measures the TSX-listed names on Bitcoin per share.
The Canadian holder’s October was worse
In Canadian dollars Bitcoin fell 8.99% this month against 7.28% in US dollars. The currency added to the loss rather than cushioning it, because the Canadian dollar strengthened 1.56% against the US dollar, from 0.7047 to 0.7157, which accounts for about 1.4 points of the difference. Ethereum is down 17.87% in Canadian dollars against 16.33% in US dollars.
The Canadian-dollar figure is also the one that matters at tax time, because an adjusted cost base is fixed in Canadian dollars at the exchange rate on the day of acquisition, not at today’s rate.
What the record says about November, and why it says very little
Bitcoin’s twelve Novembers average +7.51% with a median of +7.58%, seven of twelve positive, the best +58.21% in 2017 and the worst -36.41% in 2018.
Split them by what the preceding October did and the groups look nothing alike. After a green October, nine cases, the median November was +8.78% and six of nine were positive. After a red October there are three cases: +11.74% in 2014, -36.41% in 2018 and -17.49% in 2025. The median is -17.49% and one of the three was positive.
Now hold that against everything above. Three observations is not a base rate. This piece has just spent most of its length showing that thirteen Octobers were thin enough for one unfinished month to move the mean by nearly two points, and this split rests on less than a quarter of that evidence. The median of three numbers is only the middle one, so -17.49% is not a central tendency. It is a single November, 2025’s, standing in for a distribution that does not exist. Two of the three were sharply negative, which is worth knowing in the way a coin landing tails twice is worth knowing. Nobody should position on an n of three, and that includes us.
The regime reading cuts the other way, with a caveat. Bitcoin’s 50-day moving average has been above its 200-day since September 8, 2026, and that regime is currently 1.2% below where it began after a 10.9% drawdown. Across the 3,653 sessions since October 31, 2016 there have been nine of these golden crosses and eight completed ones. Those eight ended at a median of +12.2%, the best +292.2% and the worst -30.6%, with three of the eight ending negative, and the median regime drew down 32.4% at some point along the way. The median peak gain inside a regime was +46.0%, but a peak is a price nobody was obliged to sell at. That is eight past regimes, not this one, which has delivered nothing so far. The TradingView BTCUSD chart is there for anyone who wants to read the levels themselves.
The test worth running on any seasonal
The point is not that Uptober is dead. It is that a twelve-observation base rate was never the kind of evidence it was being quoted as, and finding that out is cheap. Check whether the pattern survives a second asset, a rival month, one new data point, and the question of how many candidates it beat to get your attention. October came through none of them cleanly: absent from Ethereum’s mean and thin on its median, beaten by February on two measures of three, two points lighter after one unfinished month, and a gap that chance produces about a third of the time once you count the eleven months it was picked ahead of.
Run that test on the next seasonal anyone hands you.
Figures are our arithmetic on Yahoo Finance daily closes, data as of 8:07 a.m. ET on October 31, 2026, with October running to a live bar rather than a close.
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. Every price in this article comes from one pull of Yahoo Finance daily closes at 8:07 a.m. ET on October 31, 2026. Bitcoin’s calendar month does not close until midnight UTC, so the October 31 figures are a live bar rather than a close. The monthly and quarterly returns, the month-of-year table, the permutation tests and the November splits are our own arithmetic on the full Bitcoin and Ethereum daily price histories from Yahoo Finance, which begin on September 17, 2014 and November 9, 2017. The cross-asset table runs September 30 close to October 30 close for every series, the last session all of them share.
