Bills for duration
Thursday’s 10-to-20-year operation took $5.187 billion against a $6 billion max. Farley says they pay with bills. Yields rose anyway. Whether that is a midterm kitchen-sink or a multi-year print is still the fight.
On August 19 the Treasury published the size. Liquidity-support buybacks of longer-dated nominal coupons — the 10-to-20-year sector and the 20-to-30-year sector — would at least double. The current maximum of $2 billion per operation becomes at least $4 billion. Effective September 9. In force through November 4, when the next Quarterly Refunding Announcement decides the size again. The stated reason is plumbing: a significant volume of high-quality offers. It is not yield-curve control by name.
The first doubled-cap operations are dated. September 10 for the 10-to-20-year sector. September 24 for the 20-to-30. On August 18, before the cap moved, dealers offered about $20 billion into a 20-to-30-year operation. Treasury took the full $2 billion. The market went higher in yield anyway.
Jack Farley, hosting Forward Guidance the next morning, did a second piece of arithmetic that is not in the press release. He said they fund the buybacks with bills. “If you're issuing treasury bills in no duration and you're buying duration, you're taking duration out of the market. If that sounds like a different variation of qe, it's because it is.” He called it fiscal Operation Twist — removing duration the public holds and replacing it with money-like bills, stimulative especially while coupon auctions are not being increased. The wiki leaves that funding claim on the podcast. The filing confirms size, sectors, and dates.
We're already living in yield curve control. When we look back 50 years from now and they're writing books about this moment in time, they will be calling what's happening right now. Yield curve control.
Darius Dale, Macro Voices, August 20, 2026
Who led the tape
Gold was up three and a half to 4 percent and “breaking out almost 4,500.” Bitcoin squeezed $1.27 billion of shorts. The dollar was down 75 basis points. Long-end yields fell. SPY was up 30 basis points. The Qs were down 10. Farley noted that dollar weakness is a headwind for foreign holders of US equities because of FX hedging costs. The vol-desk story is about keeping Oracle-class credit funded. This page is about the rotation that showed up when the sized tool landed: gold, Bitcoin, energy, short dollar — not Nasdaq.
August 20 cross-asset reaction
Jack Farley, Forward Guidance, August 20, 2026. One morning’s tape, not a filing.
Why August
About two and a half months to the midterm election. The Forward Guidance table read the August stack — yen intervention, repo-facility news, an off-cycle QRA, then this buyback — as choosing the inflation and debasement path now. They will not let the bond market lay down the law before the vote. Farley’s runway is short: high confidence through November, “nothing really happens until February,” then ugly first-quarter 2027 inflation prints against a deadlock Congress. An unnamed co-host sketched the exit: if Democrats take the House, Bessent may want the bond market to go so a split Congress is forced to cut spending. The doubled cap is written through November 4. That is a quarter-window tool as published — consistent with the kitchen-sink-into-the-midterms side, not a close on Dale’s multi-year print. The November 4 refunding is the next observation.
Inning one
The Fed-matching-the-bills leg has not happened. Farley’s extrapolation: Treasury issues $10 or $20 billion a month of bills to buy back long bonds, then the Fed matches those bills — “that's literally debt monetization.” Both hosts called August 20 inning one. Ceresna’s test is simpler: it is not QE until the Fed is the residual buyer of the new bills. Patrick Ceresna’s trade of the week that morning was IBIT around $39, long the 15-Jan-2027 $32 call at $8.75 as a high-delta stock replacement.
A late-August note put a second actor on the same stretch of curve. Treasury doubled the long-end buybacks on August 19. A Kevin Warsh task force is examining a shorter Fed portfolio — Operation Twist run backwards against the buyback window. The size and calendar of the Treasury operations are still the confirmed fact, from the August 19 release. Funding those buybacks with bills is still Farley on a podcast. The collision is unpriced, not a new gold or Bitcoin print.
Farley, the morning before Jackson Hole, quoted Stan Druckenmiller’s op-ed onto the same tool. “The long term treasury yield is the most important price in the world. It is also the only fiscal disciplinarian the US has left. Every basis point of artificial yield suppression is a subsidy to procrastination.” Farley’s own read of the US long-end move was term premium, not rate expectations. Talk that the Treasury General Account — about a trillion dollars — could fund larger buybacks is talk, not executed size. Daniel Lacalle, on Macro Voices the same stretch, said governments have already exceeded the economic, fiscal, and inflationary limits that give their debt credibility as a reserve of value. Gold is the central-bank reserve that does not threaten a domestic currency; the dollar stays the fiat reserve “because the alternatives are worse.” That is hard-asset color on the same basket. It is not a confirmation that bills fund the buybacks.
Warsh delivered the Jackson Hole keynote. He did not mention Treasury Secretary Scott Bessent’s long-end buybacks. Silence is not a Step-2 flip. Druckenmiller versus Bessent remains a Treasury story, not a Fed-chair confirmation. Size and calendar are still the August 19 release. Bills as the funding source are still Farley. Do not emit gold or Bitcoin off the missing sentence.
The All-In hosts, first on an incomplete clip and then on the speaker-labeled full episode, put more color on the same August 19 double and did not rewrite it. Calacanis had the 30-year at a 19-year high of 5.3 percent, Bessent doubling long-bond buybacks from $2 billion to $4 billion, CNBC saying Treasury was considering more, and Fox putting “the fear of God” into shorts — named outlets, not fetched pages. His Druckenmiller paraphrase is that it is wrong to manipulate prices and that America has a spending problem. That is not Farley’s quote. David Friedberg called Bessent, Druckenmiller, and Warsh a “triumvirate” of people who worked together — host color on proximity, not a confirmation that Warsh will back the buybacks. In the same stretch Friedberg had the 30-year at 5.2 percent. The hosts do not agree with each other. The press release is still the fact.
Matt Hougan, on Forward Guidance the morning of September 3, said the regime has already moved — from the Fed and QE to “the treasury and it’s more the debt that are primary drivers.” Bessent’s Treasury-market steps, in his telling, are the macro factor pulling gold and Bitcoin together. Bob Haber, on the same tape, called $12 trillion a year of rollover, T-bill-end issuance, and yield-curve twists “fiscal dominance” and “band aids.” He talks twists and bills as government strategy. He does not confirm that this buyback is funded with bills. The first doubled operation is still September 10 in the 10-to-20-year sector. The program is effective September 9. Size and calendar are still the August 19 release.
Monday morning, September 7, spot gold printed $4,405.47 — down half a percent after Friday’s one-percent slip, on a CNBC tape the wiki treats as tape, not a chain flip. Treasury’s Q3 schedule listed the announcement for September 9 — preliminary CUSIPs at 11:00 a.m. Eastern — and the first doubled operation the next day, from 1:40 to 2:00 p.m. in the 10-to-20-year sector, settle September 11, the same morning as PPI. Reuters, after that announcement, said Treasury would buy up to $6 billion of 10- to 20-year bonds in the September 10 window — triple the last long-dated operation, and above the August “at least $4 billion” doubled cap. A $6 billion maximum is not a take.
The take printed. Thursday’s 10-to-20-year operation accepted $5.187 billion against $10.489 billion offered and a $6 billion max — bid-to-cover 2.02. The largest takes were November 2040, at $1.703 billion, and May 2042, at $1.294 billion. That is above the $4 billion falsifier. A take below the maximum is not a miss of the doubled program. Yields rose anyway. The 30-year sat around 5.36 percent, highest since June 2007 — the same “market went higher in yield anyway” pattern as the August 18 $2 billion operation.
Farley, on Forward Guidance recorded that Thursday — after the $6 billion maximum printed and before CPI — said Bessent had come in above the $4 billion minimum at $6 billion. The take on the page is still $5.187 billion. His 10-year versus nominal-GDP sketch sat around 5.8 percent fair value: why yields want to go higher while Treasury tries to put a lid on them. Jens Nordvig, on The Compound the same stretch, said Bessent “is clearly willing to intervene” in both dollar-yen and the long end, and that they do not want those bond yields going any higher. The 30-year was meaningfully above the 10-year. He does not confirm that this buyback is funded with bills. A later pass put the funding mechanics in English a journalist can use: unlike the Fed, Treasury cannot create money to finance the purchases. The buybacks have to be paid with more issuance — most likely bills — or with cash from the Treasury General Account. That is still not the August 19 filing. The Fed-matching-the-bills leg stays open. As of the last wiki pass, September 24 in the 20-to-30-year sector was still the next machine check. That take is not on the page. Do not emit gold or Bitcoin off color on a take that already cleared the doubled cap.