It trades 92.01 in the premarket against yesterday’s 92.23 close. The rising channel drawn from the late-2022 low near 40 is fully intact, price broke to new highs above the 88.9 shelf this summer, and the sessions since have been a consolidation sitting above the breakout zone rather than a return into it.
On its own terms that is about as orderly as a four-year trend gets, and nothing on it records the sovereign debt problem building underneath. Benjamin Graham’s old line is the one that fits: the stock market is a voting machine, the bond market is a weighing machine. This morning they disagree about Europe.
The vote is that chart. The weighing is the long end. And between them sits a third thing that most dollar-based holders of that chart have not seen at all — because for a week now Europe has been red in its own currency and flat in theirs.
CAC 40 8,453 · 7 red sessions · 5d −2.28%
IBEX 35 19,811 · 7 red · 5d −1.77%
Euro Stoxx 50 6,422 · 5 red · 5d −1.89%
DAX 25,983 · 4 red · 5d −1.20%
HEDJ (currency-hedged Europe) · 4 red · 5d −2.17%
VGK (USD, unhedged) · 1 red · 5d −0.40% · EWU 5d +0.58% · close-to-close through 2026-08-20
Those are close-to-close counts through yesterday’s European finish. Seven consecutive red sessions in Paris, seven in Madrid, five on the Euro Stoxx 50, four in Frankfurt. HEDJ — the wrapper that strips the euro out and leaves the equities — is four red and −2.17% over five days, which is the local truth expressed in a US listing. Now the other side of the same week: VGK is −0.40% over five days with a red streak of one, and EWU, the UK fund, is actually up 0.58%. Same companies, opposite sign.
The difference is the euro, and the euro’s strength is the dollar’s weakness. EURUSD near 1.17 absorbed a week of local selling and handed dollar-based holders a flat tape while the underlying market bled. This is the house doctrine stated as plainly as it ever gets: a USD country fund is equity plus currency, and when the two legs move against each other the wrapper reports the net and says nothing about which leg did the work. Our country-flow ranking exists to hold those legs apart. Today that is not a methodological footnote. It is the story.
The weighing machine
Which brings us to the weighing machine, and to a board we are publishing for the first time this morning. Our new G7 sovereign pressure board reads the seven government curves as one instrument and scores the pressure sitting on them.
Sovereign Pressure Index +0.92 (21 sessions ago 1.48 · window max 2.31) · pressure building
z-slope +2.2 · z30 +1.2 · z10 +0.23
G7 10-year 4.015% (+8bp over 21 sessions) · G7 30-year 4.745% (+19bp) · 10s30s 73bp
Pressure building, and the components name the leader without ambiguity. The slope z-score is +2.2 and the 30-year z-score +1.2, while the 10-year z-score is only +0.23. That is a long end leading a curve, not a front end leading a policy repricing — term premium and fiscal supply rather than rate expectations. The index at +0.92 sits below the 1.48 of twenty-one sessions ago and well under the 2.31 window maximum, so the level is not an extreme and we are not going to dress it up as one. It is the direction of the components that carries the weight.
France 10y 4.11% · Italy 10y 4.06% · Germany 10y 3.27%
UK 10y 5.07% · UK 30y 5.80%
US 10y 4.702% · US 30y 5.244% · Japan 30y 4.07%
Two of those rows deserve to be read out loud. France’s 10-year at 4.11% now trades above Italy’s at 4.06% — the risk ranking that held for the whole post-crisis era, inverted, without an announcement. And Britain pays 5.07% at ten years and 5.80% at thirty. Germany at 3.27% is the anchor the bloc is measured against, and the spread from Berlin to Paris has quietly stopped being a periphery story and become a core one. The VGK chart contains every one of these countries and shows none of it.
Two bond markets to watch from here, and they are not the two the headlines pick. France is the first: the 10-year through Italy’s, and seven straight red sessions in Paris equities underneath it. Those are one repricing seen from two windows — the market adjusting its view of France’s position rather than of Europe’s. Japan is the second: the 30-year at 4.07% is the highest of the era, and the yen at 158.82 sits against the wires we have carried since Wednesday. A long end at generational highs and a currency at levels that have drawn official attention before is a combination that resolves somewhere.
The net ledger
And here the board departs from the standard ledger, deliberately. Most sovereign commentary quotes gross debt to GDP and stops there. The more honest arithmetic is debt minus assets, read against GDP — and doing it changes the map. Count Italy’s gold against its debt and the Italian position looks nothing like its headline: the Banca d’Italia sits on one of the world’s largest official gold stocks, which means every leg higher in the gold price mechanically strengthens the net position of the gold-holding sovereigns. The gold bull market is quietly recapitalising them. Net out Japan’s government financial assets and the IMF’s own ledger takes Japan from roughly 233% gross debt to GDP to about 133% net — the assets remove about a hundred points of GDP from the headline everyone quotes. And then the second dimension applies: roughly 88% of Japanese government debt is held at home, and for longer-term JGBs excluding bills the foreign share is nearer 6%. The principle we work from is this: what decides sovereign stress is the net position, read against the domestic absorption situation and the capability behind it, not the gross headline. Which is why France occupies this board more than Italy does. The net trend and the dependence on foreign holders are what that yield inversion is pricing, and neither of them appears in the ratio everyone quotes.
Italy gross 135% of GDP · ~122% after official gold alone · 2,452t (~€285bn)
France gross ~113% · 104.7% on INSEE’s net measure (2024, liquid assets — central-bank gold not consolidated) · 2,437t gold
Japan gross ~233% · ~133% net (IMF) · ~88% domestically held · long JGBs ~6% foreign
Run the clean figures and the map redraws itself. Italy: 135% gross, roughly 122% after deducting the official gold alone — 2,452 tonnes recently valued near €285 billion. France: about 113% gross and 104.7% on its own published net measure, which nets out deposits, loans and securities but does not consolidate the central bank’s gold — and France holds a nearly identical 2,437 tonnes. Italy still carries the heavier balance sheet, but its gold stock closes part of the apparent gap, and every leg higher in the metal closes a little more of it. The distance between Rome and Paris on the honest ledger is far smaller than the one in the headlines — which is precisely what a bond market that prices France over Italy has already worked out. A second part of this ledger is still to be written: accrued pension promises and the other off-balance-sheet obligations, which raise every apparent figure enormously — and which are not directly comparable with bond debt, because future contributions and the state’s power to change benefits sit on the other side of that scale. We will take that up in its own edition.
The American row is the one that moved this week, and the sequence is the point. Wednesday the Treasury doubled its long-end buyback operations. Thursday yields reversed back up through them within hours. Overnight, Bessent flagged scope for larger buybacks still, alongside a fiscal plan. The 30-year prints 5.244% and the 10-year 4.702% — back above the level the doubling was meant to defend. We wrote yesterday that the hard-asset read weakens if the 10-year backs up through the buyback. It did, inside a day. What the market is pricing is a buyback as a tool that recurs rather than a level that cleared, and a tool that has to recur is a tool that has not finished working.
The stress signature
SPY −0.84% · DIA −1.27% · IWM −1.34% · QQQ −0.72%
VIX 16.01 +7.52% · VXN 23.26 +5.54% · VVIX 89.86 · MOVE 73.18 +2.69%
GLD +0.34% · TLT −0.82% · 2026-08-20 close
Yesterday’s US session carried the same signature from the other end. Every capitalisation band closed lower and the volatility complex rose together — equity vol and bond vol on the same day, with VIX up 7.52% and MOVE up 2.69%, and VVIX at 89.86 well above its roughly 80 baseline. That combination is a cross-asset event rather than a routine equity dip. Gold up and long duration sold completes it: the bid went into things that do not have a coupon.
The count arc we drew yesterday extends into a second session. Bitcoin trades 76,960 this morning, up 5.4%, after IBIT confirmed its channel break on the close at 41.20 and the listed crypto complex broke a year of structure on Thursday. Gold futures at 4,620.8 are up 1.1%. Both readings we hold on gold still have road left and we are still not required to choose between them. The falsifier panel updates rather than closes: the 10-year has already backed up through the buyback, which was the first condition we wrote down; dollar-yen at 158.82 has taken out the 158.5 wire from Wednesday’s edition, and 159.69, the second wire, still holds. Two of the four written-down conditions are live and the hard assets are higher anyway. That is worth recording, not explaining away.
Set those two against the bond board and a shape appears that we would rather name now than after the fact. Bull markets in both hedge instruments at once — gold at highs, bitcoin’s break confirmed and extending into a second session — running alongside sovereign pressure that is long-end-led is not three stories. It is one message priced three ways: the market is charging more to hold government promises and paying more to hold the things no government issues. That is the ingredient list for what may become a perfect storm, and the timing is not neutral either, because we are entering the seasonally critical autumn window. September and October are where bond-market stress episodes have historically clustered. Seasonality is timing context and not a mechanism — it explains nothing on its own and we do not trade a calendar. But a fragile long end arriving in the window where fragile long ends have tended to break is worth having written down in August rather than in October.
The overnight tape is constructive where it has been constructive all week. KOSPI +1.29% is a second confirming session for the Korean repair, with SK hynix at ₩1,733,000 and Samsung +4.6%; the Nikkei gave back 0.60%. Nasdaq futures sit near 29,410, firmer. Today’s Morning 10 carries the overnight detail.
Put the week in one line: in local currency Europe has been selling for seven sessions, in dollars it has barely moved, and the G7 long end has made new cycle highs while the equity chart holds its breakout. The wrapper is not lying — it is reporting the net of two legs, and the leg doing the work this week was the currency. That is the whole divergence. The voting machine counts prices, in whatever currency the buyer happens to hold; the weighing machine counts duration, in every currency at once.
The scoreboard and the falsifiers
Today’s close does more than end a session. It grades a week, and four of our open cards get their verdict on it.
IEF at 93.04 — the weekly reclaim, second attempt, after yesterday’s 93.00 close. A Friday close above it would be the first weekly reclaim of the level in this leg; a second failure leaves the long end in charge of the week.
SPMO against SPLV — momentum took its first day-win in four on Thursday. The weekly verdict is today’s close, not Thursday’s.
The ex-tech pairs — QQXT against QQQ and SPXT against SPY confirmed two of three days and missed on day three. Today decides whether the rotation carries a full week or was a three-day guest.
Print records and the macro — ADI, FN and BIDU windows are open on the print record, and US flash PMIs land at 13:45 UTC.
What would break this read is worth writing down before the close writes it for us. The read is that a four-year channel survives a bond market repricing duration, and the condition it rests on has two legs rather than one. The first is the rise in long-duration yields stopping: if the G7 30-year turns down from 4.745% and the sovereign pressure index rolls back toward zero, the divergence resolves in the voting machine’s favour and Europe’s consolidation above 88.9 is a base rather than a top — that is the “all depends” in this morning’s chart note. The second leg is the short end staying low. A degree of bear steepening is tolerable, and tolerable is what we currently have: a 10-year z-score of +0.23 against a slope z-score of +2.2 and 10s30s at 73bp is a long end moving while the front end stays anchored. The version that is not tolerable is the long end running while the short end lifts with it, or the steepening accelerating from here — that is where duration repricing stops being a term-premium story and becomes a funding one. A bit of bear steepening is fine. We watch it anyway. If instead VGK loses the 88.9 breakout shelf and closes back inside the old range, the weighing machine has already won the argument and the local streaks were the leading indicator rather than the noise. A green CAC today, the first in eight, resets the count and takes the urgency out of the question for a week. And dollar-yen through 159.69 — 158.82 has already taken out the first wire at 158.5, the second still holds — would say the dollar leg is turning, which is precisely the mechanism that has been hiding Europe’s local tape from dollar-based holders in the first place.
We hold both sides of this, because both are on the screen. The four-year channel is intact and the break to new highs has kept its breakout zone: real. Seven red days in Paris and four in Frankfurt: real. The G7 long end at new cycle highs with France trading through Italy: real. What we are not entitled to is the resolution. The voting machine is the faster of the two and the weighing machine is the slower, and the reason Graham’s line has outlived its author is the order in which they finish, not the speed at which they start. All of it depends on whether the rise in long-duration yields stops soon and on the short end staying anchored while it does — a question this diary can watch, size and falsify, but cannot answer. Probability, not prophecy.