Japan's ten-year government bond yield closed at 3% on Tuesday for the first time since September 1996. The US ten-year Treasury closed at 4.79%, its highest since January 2025, and touched 4.81% overnight. Brent crude is above $95 after the United States and Iran exchanged strikes at the weekend. Gold has lost 7.3% since 24 August. The Nikkei is down 2.6% and the Kospi 3.6% this morning. That is the mainstream front page, and none of it is a bitcoin story. Bitcoin's contribution to it is that at 06:11 UTC on Wednesday 2 September it was trading at $77,656.77 on Bitstamp, 0.23% below where it closed on Friday, before any of this happened.
This desk's standing position since July is that geopolitics reaches bitcoin through the rates market or not at all, and that the right question on a morning like this is not "is bitcoin a hedge" but "which part of the yield move is doing the work". So we pulled the two Treasury series that answer it. The finding is that Friday and Tuesday were different kinds of day: the Warsh selloff on 28 August was almost entirely a real-rate move, and Tuesday's oil-driven selloff was almost entirely an inflation-expectations move. Bitcoin held through both. Gold held through neither.
What happened, in order, with clocks
On Sunday the US struck Iranian positions on Larak island in the Strait of Hormuz; Iran retaliated against US targets in the region, and on Monday Donald Trump told a Fox News reporter "we're going to hit them hard", per AFP reporting carried by Yahoo Finance. Brent futures went from $89.31 at Friday's close to $95.62 at 06:11 UTC Wednesday, +7.07%; WTI from $83.40 to $90.78, +8.85% (Yahoo Finance). The strait itself has been closed for months; what changed was the probability of the closure becoming a shooting war again.
On Monday US Treasury Secretary Scott Bessent told CNBC he expected Tokyo and the Bank of Japan to act to strengthen the yen, saying "I have information that the market doesn't have", per Reuters. On Tuesday the ten-year JGB rose 6bp to 3.00%, a level Bloomberg and Reuters both date to September 1996; Reuters puts the market-implied odds of a BOJ hike at its 17–18 September meeting at 73%. Japan's fiscal 2026 budget assumed a 3% long-term rate for debt-service costs, so this is the line above which the arithmetic in Tokyo's own documents stops working.
The two together pushed every long-dated government bond in the advanced world lower. The US ten-year constant-maturity yield closed Tuesday at 4.79%, the thirty-year at 5.27%, the two-year at 4.39%. The S&P 500 closed at 7,631.47, its third consecutive decline (7,730.99 on 27 August, 7,711.76, 7,686.14, 7,631.47). The Nikkei was at 64,477, −2.63%, and the Kospi at 6,588.15, −3.62%, both as of the same snapshot (Yahoo Finance).
"Highest since 2023" is not what the daily series says
Two morning pieces we read described the ten-year as at its highest since 2023, or "in about three years". We checked against the Treasury's own daily par-yield table for 2023, 2024, 2025 and 2026 — 917 rows, one per business day. The last close at or above 4.79% was 13 January 2025, when the ten-year closed at exactly 4.79% (it was 4.78% the next day and 4.61% three days later). The last close at or above Tuesday's two-year of 4.39% was 13 January 2025 as well, at 4.40%. So on the series that settles the question, Tuesday was the highest close in twenty months, not three years. The 2023 comparison is a claim about intraday prints, and the ten-year's 2023 high on the daily series was 4.98% on 19 October 2023, which nothing this week has approached. "Highest since" claims need a series named next to them; this one was not.
Friday was a real-rate day. Tuesday was an inflation day.
The Treasury publishes two ten-year yields every business day: the nominal par yield and the real par yield from the inflation-protected (TIPS) curve. The difference is the market's implied inflation compensation, the breakeven. Splitting each day's move into the two components tells you what the bond market was actually repricing. Here is the pull, four rows, one per business day from the close before Warsh's speech to Tuesday, plus the window total:
| Close | 10y nominal | 10y real | 10y breakeven | Day: nominal / real / BE |
|---|---|---|---|---|
| Thu 27 Aug | 4.67% | 2.34% | 2.33% | — |
| Fri 28 Aug (Warsh) | 4.73% | 2.42% | 2.31% | +6 / +8 / −2 bp |
| Mon 31 Aug | 4.75% | 2.44% | 2.31% | +2 / +2 / 0 bp |
| Tue 1 Sep (oil, Japan) | 4.79% | 2.44% | 2.35% | +4 / 0 / +4 bp |
| 27 Aug → 1 Sep | +12 bp | +10 bp | +2 bp | — |
The Warsh day was real rates: +8bp real, breakeven actually down 2bp. A central banker saying policy may not be restrictive enough raises the expected path of real policy rates and, if anything, lowers expected inflation. Tuesday was the opposite: the real yield did not move a single basis point and the breakeven rose 4bp. That is the oil signature — the market raised its inflation compensation and left the real rate alone. At the five-year, where oil matters more, Tuesday was +6bp nominal, 0bp real, +6bp breakeven. The same split at the thirty-year: +2bp nominal, −1bp real, +3bp breakeven.
Why it matters for bitcoin is that the two components act on it through different channels. The real yield is the opportunity cost of holding an asset that pays nothing; that is the channel that hurt bitcoin through 2022 and that a hawkish Fed operates. The breakeven is the debasement story; a rising breakeven with a flat real yield is, on paper, the environment in which a fixed-supply asset is supposed to do well. Friday was the first kind of day and bitcoin fell 3.04% (Bitstamp 80,278.98 → 77,835.18). Tuesday was the second kind of day and bitcoin fell 1.49% (78,571.17 → 77,397.43), with the whole of the Asian equity selloff still to come. Neither move is large next to a $6 oil move or a thirty-year JGB milestone; both are in the direction a risk asset moves, not the direction a hedge moves. Our field guide on reading the long end covers the third component, term premium, which neither series isolates.
Gold did not survive the week that bitcoin did
The hard-asset reading of the selloff — that yields are rising on fiscal fear rather than growth, and that money should therefore rotate into assets outside the fiat system — has to explain gold. It cannot. The PAX Gold token on Binance, which tracks an ounce of gold around the clock, closed at $4,673.67 on 24 August and was at $4,333.04 at 06:11 UTC Wednesday: −7.29% since the 24 August close, of which −2.54% was Tuesday alone. CME December gold futures went from $4,609.70 on 27 August to $4,348.00 at Tuesday's close, −5.68% (Yahoo Finance); Reuters quoted spot at $4,393.11 on Tuesday, and CoinDesk's summary is that gold has come "from $4,700 per ounce to $4,300 per ounce in less than a week".
Over the same window bitcoin, on Bitstamp closes, went from $78,986.06 to $77,656.77: −1.68%. From Friday's close it is −0.23%. From the 27 August close, the day before Warsh, it is −3.27%. Whichever window you pick, gold has fallen roughly two to four times as far as bitcoin during a stretch in which real yields rose 10bp and the dollar index firmed to 99.70. That is not evidence that bitcoin is a hedge; the sign of its moves is wrong for that. It is evidence that in this particular selloff the asset that is usually sold first on a real-yield shock is gold, and the asset with a $54.7 billion cumulative ETF bid behind it has a floor gold currently lacks. It is worth remembering that the same ETF bid ran negative on Tuesday, −$236.5 million on Farside's completed table, the largest outflow since 31 July — we grade that separately in today's Markers.
What bitcoin actually did
| Bitstamp close (UTC) | Close | Day | High | Low |
|---|---|---|---|---|
| Thu 27 Aug | $80,278.98 | +1.59% | $80,819.98 | $78,573.39 |
| Fri 28 Aug (Warsh) | $77,835.18 | −3.04% | $81,454.98 | $76,876.88 |
| Mon 31 Aug | $78,571.17 | +1.12% | $79,249.85 | $77,381.47 |
| Tue 1 Sep (oil, Japan 3%) | $77,397.43 | −1.49% | $79,184.00 | $76,380.13 |
| Wed 2 Sep, 06:11 UTC | $77,656.77 | +0.34% so far | $77,747.28 | $76,682.57 |
Five rows, five days, no gaps. Tuesday's low of $76,380.13 was set during the US session and is the lowest print on Bitstamp since 23 August; the overnight Asian selloff, which took the Kospi down 3.6%, did not retest it — Wednesday's low so far is $76,682.57. The 24-hour range on Bitstamp at the snapshot was $76,380.13 to $78,994.68, −1.66% on the day. CoinDesk's Shaurya Malwa noted that solana and tron each fell more than 3% while bitcoin gave up about 1%: the high-beta names were sold first and the base layer left alone, which is what a de-risking looks like from inside crypto.
The one thing in the picture that is a genuine headwind is the dollar. The dollar index was at 99.70 at the snapshot, up from 99.16 on 27 August, and CoinDesk's Omkar Godbole points out it is sitting on a trendline from the 2011 lows that traders watch. Bitcoin's inverse relationship with the dollar is looser than the one with real yields but it is real, and a dollar bounce on a Japanese intervention would be a dollar bounce.
The three dates that settle it
Friday 4 September, 08:30 ET: August payrolls, consensus roughly +55,000 after July's −23,000. Tuesday's JOLTS already leaned soft — openings 7.271 million against 7.33 million expected, hires down 278,000 to 5.054 million, the quits rate at 1.9% — and ISM manufacturing came in at 54.6 against 55.2 expected, with prices paid at 71.1. A hot payrolls number hardens the hike case that CME FedWatch had at 66% at 05:01 UTC per CoinDesk (66.4% on Tuesday per Yahoo Finance), up from roughly 36–40% before Jackson Hole. Wednesday 16 September, 14:00 ET: the FOMC decision.Thursday 17 to Friday 18 September: the Bank of Japan, which markets now price at 73% for a hike. A BOJ hike is the one event on the calendar that raises yields in Tokyo and, through the carry trade, could lower them in New York; it is also the one event that strengthens the yen and weakens the dollar. Which of those dominates for bitcoin is not knowable in advance, which is why it is a marker below and not a forecast.
The markers
Two falsifiable claims, as standing practice, each naming its settlement source in the same sentence as its bar. I1: the ten-year real yield on the Treasury's daily real par curve closes at or above 2.50% on any business day through Wednesday 16 September 2026. The 2026 high is 2.47% on 31 July; Tuesday closed at 2.44%. If the selloff is the real-rate story the Warsh day suggests, this bar falls; if it is the oil-and-breakeven story Tuesday suggests, it does not. I2: the Bank of Japan raises its policy rate at the 17–18 September meeting, settled on the BOJ's own statement. The market says 73%; Bessent says he has information the market does not. One of them is wrong, and the statement will say which. Both join yesterday's H3 (thirty-year CMT at or above 5.25% on 17 September; Tuesday's close 5.27%) on the board.
Method: bitcoin prices are Bitstamp BTC/USD daily candles (UTC) and Binance BTCUSDT; open interest, funding and long/short ratios are Binance USDT-M and COIN-M public endpoints; Treasury yields are the US Treasury's daily par nominal and real constant-maturity series; ETF flows are Farside Investors' completed daily table; gold is the PAX Gold (PAXG) daily candle on Binance as a 24-hour proxy and CME gold futures via Yahoo Finance; equities, oil and the dollar index are Yahoo Finance quotes. Every figure labelled with today's date was snapshotted at 06:11 UTC on Wednesday 2 September 2026 unless a different clock is printed beside it. Windowed pulls carry their row counts.
Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrencies are volatile and you can lose money. Nothing here is a recommendation to buy or sell any security, token or exchange-traded fund. Do your own research and consult a licensed financial advisor before making investment decisions.