On Thursday SPY fell 4.41%. Bitcoin, the asset that was supposed to be digital gold, fell 5.55% the same day. That surprises nobody who has watched crypto since 2020, but the two stories still get told side by side: BTC is a hedge against a falling dollar and a shaky system, and BTC is just a high-beta tech stock that sells off with everything else.
We checked every day SPY closed down 3% or more since BTC price data begins in September 2014: 36 days, 17 separate selloff episodes. The answer depends on the year. Before 2020, BTC ignored stock crashes: it closed higher on 55.6% of those days, and its daily correlation with SPY was 0.02, effectively zero. Since 2020 it has fallen with them: up on only 25.9% of crash days, median same-day move -3.2%, correlation 0.39. And the next 20 days have not reliably delivered a bounce either. The median 20-day return after a crash day was +0.01%, against +1.57% on any US trading day.
What we measured
Data: yfinance SPY (adjusted) and BTC-USD daily closes, 3,053 US trading days from September 2014 through this week.
- Crash day: SPY close-to-close return of -3% or worse.
- BTC same-day move: BTC's UTC daily bar for that date. That bar closes at midnight UTC, about four hours after the US close, so it covers the stock session plus the evening after.
- Entry for forward returns: BTC's close that night. You know SPY's close before then, so the entry is achievable.
- Forward windows: 1, 5 and 20 calendar days (BTC trades every day).
- Baseline: every US trading day in the same window, measured the same way.
- Clustering: crash days come in bunches (March 2020 alone had several). An episode is a crash day with no other crash day in the prior 10 trading days. 36 crash days collapse to 17 episodes. With 17 episodes, every forward-return result here is suggestive, not proven.
What the data says
BTC on the crash day itself:
| BTC up on the day | BTC median move | Daily SPY-BTC correlation | |
|---|---|---|---|
| All 36 crash days | 33.3% | -2.78% | |
| 2014-2019 (9 days) | 55.6% | +0.31% | 0.02 |
| 2020-2026 (27 days) | 25.9% | -3.2% | 0.39 |
| Any US day | 51.7% | +0.1% |
The regime change is the story. In the earlier period, BTC traded on its own news, and a stock crash told you nothing about it. On the Brexit day in 2016, SPY fell 3.59% and BTC rose 6.62%. On the August 2019 tariff selloff, SPY fell 3.01% and BTC rose 7.62%. Since 2020 the correlation has lived between 0.19 and 0.56 every year, and it was 0.47 in 2026 so far. BTC fell more than SPY on 36.1% of all crash days. In the 2020-2026 set its mean crash-day move was -5.22%, bigger than SPY's own average drop of -4.36% across all crash days.
BTC over the following days, vs every US trading day:
| Horizon | Crash days: median | Baseline: median | Crash days: up | Baseline: up |
|---|---|---|---|---|
| 1 day | -0.1% | +0.1% | 47.2% | 52% |
| 5 days | -0.65% | +0.44% | 48.6% | 54.2% |
| 20 days | +0.01% | +1.57% | 51.4% | 55.6% |
Counting each selloff once (17 episodes) the picture is weaker still: a 5-day median of -2% and up 37.5% of the time, and a 20-day median of -1.04%, up 43.8% of the time.
Two things stand out.
There is no reliable "buy the crypto dip" after a stock crash. On every horizon BTC's median after a crash day sat below its median on a normal day, and the hit rate was below baseline too. The gaps are small. With 36 days the honest reading is "no edge, slightly negative", not "sell signal".
The range of outcomes is much wider. The middle half of 20-day outcomes after a crash day ran from -14.15% to +15.27%. On a normal day it runs from -6.14% to +11.3%. The thing a stock crash predicts for BTC is not direction. It is a much bigger move in either direction.
The threshold matters, and that is a warning sign
| SPY drop | Crash days | Episodes | BTC 20-day median | BTC up after 20 days |
|---|---|---|---|---|
| -2% or worse | 99 | 41 | +0.9% | 55.1% |
| -2.5% or worse | 59 | 28 | +0.01% | 51.7% |
| -3% or worse | 36 | 17 | +0.01% | 51.4% |
| -4% or worse | 17 | 8 | +10.78% | 68.8% |
The deepest crashes, SPY -4% or worse, were followed by a +10.78% median BTC return over 20 days. That looks like a capitulation-bounce rule, but it rests on 8 episodes, and most of the days come from two events: mid-to-late March 2020 and the April 2025 tariff crash. The 2%, 2.5% and 3% cuts show no comparable edge. A result that appears only at one threshold, on 8 episodes, is a story about two specific selloffs, not a rule.
The actionable rule
Since 2020, do not count on BTC to offset an equity selloff. On a day SPY drops 3% or more, the base case is that BTC drops too, often by more. If you hold both, your real risk on a crash day is closer to one bigger position than to two diversified ones. And do not treat the next week as a dip to buy on reflex: the median 5-day BTC return after a crash day has been below a normal week's. Size for a wider range of outcomes, in both directions.
Where it fails
Ten of the 35 crash days with a full 20-day window were followed by a BTC drop of more than 10%. The worst was February 25, 2020: SPY fell 3.03%, BTC fell 3.2%, and over the next 20 days BTC lost another 46.3% into the COVID crash. The 2022 cluster (April, May, June) was the same pattern spread out: crash days in a bear market that kept leading to lower prices.
But 16 of the 35 beat BTC's normal 20-day median, and several were huge:
- February 5, 2018: SPY -4.18%, BTC -15.97% that day, then +39% over the next 20 days.
- March 12, 2020: SPY -9.57%, BTC -37.17% in one UTC day, the worst same-day move in the sample, then +32.9% over 20 days.
- April 2025 tariff crash: SPY fell 4.93% and 5.85% on back-to-back days, BTC rose 0.75% and 0.89%, and was up about 12% over the next 20 days. That is the one post-2020 crash where BTC behaved like the hedge its fans describe.
So the rule breaks in both directions. Some crashes are the start of a much bigger BTC loss, and some mark the bottom. A single crash day does not tell you which one you are in.
The newest crash day, November 5, 2026, has no forward data yet. SPY fell 4.41% and BTC fell 5.55%, in line with the post-2020 pattern.
How to watch it
What has moved BTC on these days is the equity risk regime, not anything crypto-specific. Market Health tracks that regime (breadth, VIX and trend) and flags when it turns hostile, which since 2020 has been the time BTC is most likely to trade like a leveraged stock.