What the US government pays to borrow for ten years, day by day since 2013, with bitcoin's price on the same axis of time — and what bitcoin actually did after every big move in that rate.
4.95% the yield on Oct-22 2026, the last day in this build, read from a public record, published by askthetape.com.
The ten-year yield is what the US government pays to borrow for ten years, and it is the number most people mean by «rates». On 2026-10-22 it closed at 4.95%. Over the 5,936 days this page covers it has been as low as 0.520% (2020-08-04) and as high as 5.310% (2026-10-05). «Rising yields are bearish for bitcoin» is said often enough to be worth checking, so here is the check rather than the opinion. We marked every time the yield rose by at least 50 basis points — half a percentage point — inside 60 days, counting each run once: 24 episodes since 2013. Thirty days after those days bitcoin's median move was +5.3%, and 14 of 24 were positive. Ninety days after, the median was +17.2%. Against that, a day picked at random on this page is followed by a median +3.2% over thirty days and +13.4% over ninety. We measured larger median moves after the rising-yield episodes than after an ordinary day in this sample, and this small sample does not establish a rule: 24 episodes is a small pile, and 6 of them fall in 2022 alone. The 21 episodes where the yield FELL 50 bp instead were followed by a median +10.8% over ninety days, 15 of 21 positive. Read the bars, not the headline: each one is a single episode, hollow if it is too young to be scored.
| leg | FRED series DGS10 — the 10-year US Treasury yield at constant maturity, the Board of Governors' own daily fixing from the H.15 release (16,188 rows from 1962-01-02; the value is the yield IN PERCENT, measured on this pull: 1981-09-30 = 15.84, 2020-03-09 = 0.54) × Bitstamp daily closes. The two legs are joined on a calendar-day grid: the yield is carried forward from its last fixing for up to 7 days, so a weekend or a US holiday keeps the bitcoin day rather than dropping it; a gap longer than 7 days would break the line instead. |
|---|---|
| cadence | daily; the bond market is shut on weekends and US holidays, so the yield leg prints about 5 days in 7 and the page's freshness allowance is 4 days |
| since | 2010-07-17 — the first day both legs exist (our bitcoin daily leg starts 2010-07-17; DGS10 reaches back to 1962-01-02 and is not drawn before the bitcoin leg begins) |
| credit | the yield series is FRED DGS10, served by the Federal Reserve Bank of St. Louis; the number itself is the Board of Governors of the Federal Reserve System (US), H.15 Selected Interest Rates. Both are named because both made it: FRED for the series, the Board for the fixing. |
| 🩸 switched from a public market-data ^TNX quote on 2026-09-12 | this page used to run on the data feed's ^TNX index closes. The swap was measured before it was made, on the 14,148 days the two share (1970-01-02 → 2026-09-10): median difference 0.0000 points, median ratio 1.000000, p90 |difference| 2.1 bp, p99 4.4 bp, worst day 10.9 bp (2000-02-09); 82 days of 14,148 (0.58 %) sit more than 5 bp apart, and only 3 of those fall in the last five years. Three things made DGS10 the better leg. It is DEEPER: 16,158 fixings from 1962-01-02 against 14,193 index closes from 1970-01-02. It has no HOLES: the ^TNX feed carried 102 rows (0.71 %) with an empty close that the old pull dropped without saying so. And its CALENDAR is honest: ^TNX printed a yield on 44 days when the US Treasury market was shut — every Columbus Day and Veterans Day since 1994 — while DGS10 skips those and instead covers the 16 days since 1970 when the stock exchange was shut and bonds traded (nine Good Fridays, 2001-09-13/14, Hurricane Sandy, the Ford and Carter mourning days). What it costs: the fixing is rounded to two decimals instead of three (at most half a basis point) and it publishes about one business day later, so this page's last point is a day older than it was. On this page's own window nothing moved: 4,851 days became 4,850 and the episode counts are the same, 19 up and 16 down. |
| episode rule | an episode is the FIRST day the 60-calendar-day change in the yield crosses ±50 bp after at least 30 days without one, so a long run counts once instead of ninety times. Up: 25 episodes (2010, 2011, 2013, 2015, 2016, 2018, 2021, 2022, 2023, 2024, 2026). Down: 21 episodes (2011, 2012, 2015, 2016, 2019, 2020, 2022, 2023, 2024, 2025). Forward returns are close-to-close on the bitcoin leg, 30 and 90 calendar days out; an episode younger than its horizon is drawn hollow and is not scored. |
| 🩸 the distribution, not the verdict | after the yield rose: median +5.3% at 30 days (14 of 24 positive), +17.2% at 90 days (16 of 24), worst -31.4%, best +135.9%. After it fell: median +5.2% at 30 days (12 of 21), +10.8% at 90 days (15 of 21). Every-day baseline on the same window: +3.2% at 30 days (57% positive), +13.4% at 90 days (61% positive). The samples are 25 and 21 episodes — small enough that one 2021 or one 2022 changes the median, and they are not independent draws. |
| 🩸 how little the two move together | day by day there is almost nothing to see: the 60-day change in the yield and the 60-day change in bitcoin have a correlation of -0.06 over 5,936 days. On levels, the yield against the logarithm of the price gives +0.33 — and that is two things drifting up over a decade, not a link. This page keeps both numbers printed so the chart is not read as one. |
| 🩸 the two change views are not on one axis | basis points and percent are different rulers, so they get their own view each rather than one chart with two scales, which would let the picture be tuned by choosing the scales. They share the same dates: flip between them. |
| note | the yield is a price of money set in a market we do not capture ourselves — this leg is the public FRED series, read from our own daily poll of it: if the pull fails the previous file stays and its age shows up in the freshness check |
| this build | data to Oct-22 2026 |
| policy | no backfill · no export · read it here · read on the site · not exported: this page is drawn from someone else's archive |
The chart itself is drawn in the browser: turn JavaScript on to hover a day, pick a window and draw your line. The numbers above come from the last build, Oct-23 2026 21:24Z.
What the tape says today · what's new · all 111 metrics