The standard explanation for bitcoin's recent weakness has been rising Treasury yields: when the 10-year climbs toward 5.3%, safer assets compete with risk assets. Thursday 8 October 2026 did not fit that script. The 10-year par yield fell 6 basis points to 5.22%, its biggest one-day drop of the month, according to the U.S. Treasury's daily yield data — and bitcoin still lost 1.90%, closing at $81,692.79 after an intraday low of $80,314.70 (Coinbase). So does a falling yield help bitcoin at all? This desk checked the record since 1 September.
The scorecard: seven sessions when the 10-year yield fell
Using the Treasury's constant-maturity 10-year series and Coinbase's daily BTC-USD closes (Coinbase Exchange candles), we compared each Treasury business day from 2 September to 8 October (26 sessions) with bitcoin's close-to-close change over the same interval. On seven of those sessions the 10-year yield fell. Here they are:
| Date (2026) | 10-year par yield | Change (bp) | BTC close-to-close |
|---|---|---|---|
| 3 Sep | 4.77% | −2 | +5.12% |
| 17 Sep | 4.94% | −7 | +0.27% |
| 21 Sep | 4.96% | −5 | +7.07% |
| 25 Sep | 5.17% | −1 | −0.35% |
| 1 Oct | 5.24% | −5 | +1.55% |
| 6 Oct | 5.27% | −4 | −0.24% |
| 8 Oct | 5.22% | −6 | −1.90% |
Bitcoin rose on four of the seven yield-down sessions and fell on three. The simple average is a healthy +1.65%. On the 17 sessions when the 10-year yield rose, bitcoin fell on ten and the average was −0.28%. On the face of it, that looks like a clear rule: yields down, bitcoin up.
Why that average is misleading
Two sessions do almost all the work. The +5.12% on 3 September and the +7.07% on 21 September are the two largest daily gains in the sample, and nothing in a yield series can say why bitcoin jumped on those days. Remove them and the other five yield-down sessions average −0.13% — statistically indistinguishable from the −0.28% average on yield-up days. The median tells the same story: +0.27% on yield-down days against −0.21% on yield-up days, a difference of about half a percentage point on a coin that routinely moves several percent in a day.
| Group | Sessions | BTC fell on | Mean BTC change | Median BTC change |
|---|---|---|---|---|
| 10-year yield down | 7 | 3 | +1.65% | +0.27% |
| … excluding the two largest gains | 5 | 3 | −0.13% | −0.24% |
| 10-year yield up | 17 | 10 | −0.28% | −0.21% |
| Unchanged | 2 | — | — | — |
This is a small sample and the alignment is imperfect: Treasury par yields are end-of-day New York readings, bitcoin's Coinbase day ends at midnight UTC, and Monday's bitcoin change spans the weekend. Treat the table as a sanity check on a popular narrative, not as a model. It is consistent with this desk's earlier finding that the daily correlation between yield changes and bitcoin is weak (yields at their highest since 2002 and a −0.29 daily correlation): rates matter at the margin, but they are far from the whole story.
What was driving Thursday instead
The most visible driver was equities. A Financial Times report, relayed by Reuters, that OpenAI's annualized revenue was about $20 billion below earlier investor estimates knocked the Nasdaq Composite down 1.25% while the Dow ended up 0.10% (Yahoo Finance live blog; details in our news report). Forbes quoted Blockware's John Haar on the day: “Bitcoin is down amid a broader macro risk-off move, with the Nasdaq currently down approximately 1.25%” (Forbes, 8 October). That is a cross-asset explanation rather than a rates one, and it fits the hourly record: bitcoin's worst hour, −2.03% in the 15:00 UTC hour, came during the U.S. equity session.
Video: More Crypto Online, published 8 October 2026 — a technical-analysis take on the break of support. The views are the presenter's own and are not endorsed by this desk.
The second driver was positioning. Bitcoin had rallied from about $76,000 in mid-September to briefly above $87,000 on 2 October, and CryptoQuant's Julio Moreno told Forbes that traders' unrealized profit margins reached 33% last week, the highest since December 2024, with holders realizing as much as 25,700 BTC of profit in a week, the most so far in 2026. When that many holders sit on gains, a risk-off day invites selling. CoinGlass data cited by CryptoSlate showed about $1.16 billion of liquidations in 24 hours, roughly 90% of them long positions, and short-term holders moving coins to exchanges in size, mostly at a loss.
“Traders' On-chain Unrealized Profit Margins reached 33% last week, the highest since December 2024.” — Julio Moreno, head of research, CryptoQuant, quoted by Forbes on 8 October 2026
Levels that now matter
Glassnode data cited by CryptoSlate on 8 October put the heaviest resting bids at $81,000–$81,250 on Binance's spot book and the heaviest asks at $86,500–$86,750. The first has been breached; the low of $80,314.70 sits about 0.8% below it. The same Glassnode work flagged a liquidation cluster at $81,700–$83,300 (already cleared on the way down) and another near $75,000. Bitcoin's September 15 low of $74,887.50 sits just below that second cluster. None of this is a forecast; it is a map of where leveraged positions were reported to be concentrated, and maps go stale quickly.
Three objections to this scorecard
Timing. Treasury par yields are struck near the New York close, while bitcoin trades around the clock. Much of bitcoin's Thursday move happened in the 15:00–17:00 UTC window, which is before the end of the New York session, and the Motley Fool's close report says yields fell as the session progressed after briefly touching 5.35% earlier. A daily close-to-close comparison therefore cannot tell us whether yields and bitcoin moved in opposite directions at the moment bitcoin was falling fastest.
Level versus change. The scorecard tests daily changes, but a market can be hurt by the level of rates even on days the level eases. At 5.22% the 10-year is still 45 basis points above its 3 September reading of 4.77%, and a six-basis-point retreat does little to alter that. Bitcoin is also about 8% below its 2 October intraday high of $87,249 even though yields are lower than they were on that day (5.28%).
Third variables. On Thursday at least three forces overlapped: the AI-revenue scare in equities, oil, with Brent above $104 in the morning before President Trump's comment that the U.S. would not attack Iran before the midterm elections, and a long-heavy derivatives market. A scorecard with only one explanatory variable cannot separate them, which is why this article treats the result as a caution against a single-cause story rather than as a trading rule.
What would change the picture
- A bigger sample. Seven observations cannot settle the question. Re-running the scorecard after the 14 October CPI and 28 October FOMC decision will roughly double the number of rate-driven sessions.
- A second variable. The honest model uses at least two inputs — yields and a technology-equity index — because Thursday's loss only makes sense with both. We do not have a clean Nasdaq series in this scorecard, which is its main limitation.
- A bounce that holds. Bitcoin recovered to about $82,400 early on 9 October. If yields keep slipping and the Nasdaq stabilizes, the cross-asset explanation predicts a further recovery; if the Nasdaq keeps falling while yields fall, the equity-led explanation is strengthened.
Method: 10-year yields are the U.S. Treasury's daily par yield curve (constant-maturity) values for 2 September–8 October 2026; bitcoin returns are Coinbase Exchange BTC-USD UTC daily closes, each measured against the close on the previous Treasury business day; means and medians were computed by Bitcoin Mastery on 9 October 2026. Third-party figures are attributed to their source and date; CryptoSlate's liquidation split and short-term-holder figures vary slightly inside the same article, so we quote approximate values.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. Bitcoin and other cryptocurrencies are volatile and you can lose some or all of the money you put in. Nothing here is a recommendation to buy, sell or hold bitcoin, any exchange-traded fund, any listed security or any other asset, and the technical levels, probabilities and scenarios discussed are descriptions of published data, not forecasts. Do your own research and consult a licensed financial advisor before making any investment decision.