Why bitcoin falls when US stocks sell off, and when the link breaks

The IMF sample shows a stronger link after 2020

The IMF compared the relationship before and during the pandemic. Three of its researchers reported the results in a January 2022 IMF blog post. Across 2017–19, the correlation between daily returns on bitcoin and the S&P 500 was 0.01: no visible linear relationship. Across 2020–21 it was 0.36. Against the MSCI Emerging Markets index the figure for 2020–21 was 0.34, which the authors describe as a 17-fold increase on the preceding years.

The same post estimates that bitcoin volatility explained about one-sixth of S&P 500 volatility during the pandemic, and that spillovers ran both ways. Shocks travelled from bitcoin to equities as well as from equities to bitcoin, and both directions grew markedly stronger in 2020–21.

Correlations drift. The 0.36 belongs to 2020–21 and should be read as a description of that window, not as a constant you can plug into today. What explains why the link tightened is more useful than the number itself, and the evidence points to three channels.

A fund holding both can sell both when it cuts risk

The IMF authors link stronger co-movement to wider adoption. A useful way to understand the potential overlap is to consider a multi-asset fund holding both equities and bitcoin: a decision about the whole portfolio can affect both holdings. US spot bitcoin ETFs also provide a route to bitcoin exposure through a brokerage account, although owning ETF shares and owning bitcoin directly remain different things. This mechanism alone does not show which holders sold on a particular day.

Once the holders overlap, the real question becomes what they sell first when they decide to cut overall risk. For a fund that sizes positions by volatility, a jump in measured volatility can trigger a lower exposure limit. It could sell across its holdings or sell whichever assets it can turn into cash. Bitcoin can be part of that sale because it trades around the clock. This is a possible portfolio response; the correlation figures do not reveal how often funds use it or which asset they sell first.

Trading hours matter here. US equities have extended-hours sessions, so the regular New York close does not make all stock sales impossible. When the relevant stock market is closed, however, a holder may still be able to sell bitcoin. Bad news over a weekend can therefore affect bitcoin before the next regular stock session. That timing alone does not identify the sellers or forecast Monday’s open.

Margin calls: losing money elsewhere, selling bitcoin to cover it

The second channel is more mechanical. A leveraged investor who loses money in any market can face a margin call, and brokers and exchanges tend to raise margin requirements when volatility jumps. Meeting the call may involve adding cash or eligible collateral, reducing positions, or facing liquidation under the account terms. The SEC’s margin-call explanation describes the cash, securities and forced-sale routes for brokerage accounts. An investor without spare collateral may sell another holding to raise cash; selling bitcoin is one possible response, not the only way to meet a call.

5 August 2024 shows how far that can reach. The Nikkei 225 closed down 12.4%, its biggest one-day percentage fall since 1987, according to Reuters’ closing report. The Bank for International Settlements reviewed the episode in BIS Bulletin No 90, published on 27 August. Its account: a somewhat disappointing US labour market release on 2 August acted as the catalyst, and the unwinding of leveraged positions in equity and currency markets amplified the reaction. The S&P 500 fell 1.8% that Friday and another 3.0% on the Monday. For yen-funded carry trades going into the event, the bulletin gives a rough middle ballpark of ¥40 trillion, about $250 billion, and adds that the figure is, if anything, biased down by data gaps.

Crypto appears in the bulletin too. The authors note that bitcoin and ether posted losses of up to 20% during the episode, and read that as a sign that retail traders faced margin calls and may have been forced to close positions even in seemingly unrelated assets. CoinDesk reported bitcoin briefly touching the $50,000 area that day, its lowest since February. The case supports the idea that a need for cash can carry selling from one market into another. It stops short of showing which accounts sold how much bitcoin to meet which call; the bulletin offers no trade-level evidence of that.

Crypto leverage can then amplify the initial move. If outside selling takes the price through leveraged longs’ liquidation levels, forced sales can add further pressure. The step-by-step version is in how liquidation cascades amplify moves. This gives a possible explanation for bitcoin falling further than stocks. Establishing that it happened requires evidence of both the initial selling and the subsequent liquidations.

Sometimes there is no contagion at all, just one rate headline

The first two channels describe pressure crossing from one market to another; it can start in equities, currencies or crypto. The third needs no crossing. One cause can lean on both markets at the same moment.

The usual suspect is the expected path of interest rates. A hot inflation print or a hawkish line from a Fed official pushes back the expected timing of rate cuts, Treasury yields climb, and both long-duration growth stocks and an asset that pays no yield, like bitcoin, come under pressure together. The full transmission path is laid out in Fed rates and bitcoin, and why the release minute itself gets so violent is covered in the CPI release day piece.

Telling a common cause apart from contagion changes what you watch. If rates are doing the work, the useful screens are Treasury yields and the dollar, more than the Nasdaq. If instead one heavyweight missed earnings and the rates market has barely moved, its earnings report provides a stock-specific explanation. It does not establish a direct reason for bitcoin to fall, though shared holders or wider risk reduction could still connect the two.

When bitcoin doesn’t follow stocks: one side has its own story

Mid-November 2022 is a clean example of two stories running in the same week. On 10 November the Bureau of Labor Statistics published October CPI: up 7.7% on the year, a slower pace than in earlier months. The S&P 500 closed up about 5.5%, its biggest one-day percentage gain in more than two and a half years, Reuters reported.

Bitcoin bounced that day as well, according to CoinDesk’s market wrap. The bounce came straight after a crash of its own making: as the FTX crisis deepened, bitcoin had fallen below $17,000 on 9 November. FTX filed for bankruptcy protection on 11 November, a date you can check against the Chicago Fed’s review of the 2022 runs. A macro surprise and an exchange collapse took turns at the wheel within three days.

Read the calendar first. The joint rally after CPI shows both markets reacting to the same macro news. It tells you nothing about how the FTX damage spread afterwards. Watching only the stock market would miss the withdrawal freezes, the run and the bankruptcy, and a single day of moving together is too thin to conclude that the old link had snapped back into place.

The reverse is possible too: sector-specific news can push US stocks in a different direction from bitcoin. That 0.36 coefficient cannot tell you which news or holders drove a move.

Five checks before you decide bitcoin is “following” the stock market

1. Line up the clocks. Regular trading on the New York Stock Exchange opens at 09:30 Eastern. Under NIST’s daylight saving rules, US daylight time ends on 1 November 2026, so the open is 13:30 UTC until then and 14:30 UTC afterwards. If you’re in the UK, note that British clocks go back on 25 October 2026, a week earlier, so for that one week New York opens at 13:30 London time instead of the usual 14:30. Did bitcoin’s drop start after the open, or was it already under way in Asian hours? Stock futures and pre-market trading can move first, so the opening bell alone won’t settle who led whom.
2. See what actually fell in stocks. A broad sell-off across the index, or one sector and a few big names? In the second case, look for evidence of shared holders selling before attributing bitcoin’s move to those stocks.
3. Check whether yields and the dollar moved with it. A clear move in Treasury yields and the dollar index points to a common macro cause. If both are quiet, the stock market is more likely dealing with its own problem.
4. Look for leverage being cleared in crypto. The funding rate and open interest tool provides a current snapshot from OKX. It does not supply a history of open interest or liquidation totals. To check for a sharp drop, compare observations of the same contract, unit and venue across the event, then consult liquidation records for that window. Open interest can fall because traders closed positions voluntarily or because they were liquidated, so it can’t confirm a cascade on its own. If liquidations cluster at the same time, the extra slice of bitcoin’s fall probably has a leverage component. The market heatmap shows 24-hour moves for its selected Binance pairs; it is a breadth check for that sample, not every cryptoasset or a minute-by-minute event study.
5. Last, check crypto’s own news. Exchange notices, regulatory actions, large on-chain transfers. If something big is happening here, as in November 2022, the stock market is just the backdrop.

These checks narrow the explanation down to whichever fits the evidence best: a shared macro shock, cross-market deleveraging, or a crypto-specific event. They can’t split a move into percentages by cause, and two markets falling together still doesn’t prove that stocks pushed bitcoin. When the flow or event evidence is missing, the honest label is “unconfirmed”; seeing the move happen is different from knowing why it happened.

Common questions