The Silence in the Doctrine

What this document is. An exercised scenario report: a live question played out as a game rather than argued in an essay, then written up with the record attached. I took one question about the US government's debt and its central bank, turned it into a game with seven roles, and played the game 45 times on a computer. Nobody sat at a table. Each role is a computer program that follows a written policy. No person and no chatbot plays a role. Everything on the public record up to 3 September 2026 is fact and carries a source marker. Everything after that date is the game's own record.

The question. The Treasury is the US government's finance department. It borrows what the government spends by selling bonds: IOUs that pay interest and are repaid on a set date. The Federal Reserve, the Fed, is the US central bank, and it sets the short-term interest rate. Washington has quietly agreed that the Treasury, not the Fed, decides the maturity structure of the public debt: the mix of short-dated and long-dated bonds the government owes. That agreement is the doctrine in this report's title. Nobody has agreed whether the Treasury may use that power to push long-term interest rates down. Call that the price question. A bond's price and its yield, the interest rate it pays, move in opposite directions, so holding a yield down means holding a price up, and this report uses both words for the same thing. The buyer's question, in the buyer's own words: Who settles that, and what will we see on 4 November that tells us it was settled? On 4 November, a real and scheduled date, the Treasury publishes its quarterly refunding statement: its announcement of how much debt it will sell over the next three months, and why. Its own published test for cutting bond sales is trends in structural demand, meaning a lasting shift in who wants to buy which bonds 9. So if it cuts, it will give a reason. What it says in each ending below is the game's story. Nothing there is a forecast. In this game, what the Treasury decides on 4 November is which reason to print.

The answer. My starting thesis: nobody at the Fed or the Treasury has an incentive to ask the price question out loud, so it gets decided without ever being asked. The games amended that. Across three Treasury plans played from 9 September to 4 November 2026, no speech forced the price question into the open. Only money did. The Treasury's own spending, large enough for the market to notice, was the one thing that opened the question. And the Treasury is the player with the strongest reason to keep it shut.

How it was made. Seven seats, each a program following a written policy. Five turns of one to two weeks each, from 9 September to 4 November. The run files call a turn a tick. Each turn a seat claims a move, and a referee program called the arbiter sets a difficulty number for it. A die decides the rest. The roll, plus any bonus the arbiter grants for the seat's position, must beat the number, or the move fails. The lowest possible roll, a natural 1, fails badly whatever the bonus. Every move has a written result for success and for failure. A seed is the number that fixes the dice for a whole run, so the same seed gives the same rolls. The Treasury had three plans: keep its current language, cut long-term bond sales and cite structural demand, or cut and state the yield level. I call them the keep-the-language plan, the cut-on-demand plan and the cut-on-level plan. Each plan ran on the same three seeds, the matched trio, so the plans compare move for move. Then twelve more seeds behind each plan, the ensemble. That is 15 runs per plan and 45 in all. Only the cut-on-level plan has the Treasury spend extra money, so 30 of the 45 runs are played with no extra spending.

Three scales I wrote. All run 0 to 100. All are bookkeeping, and none is a measurement. The exposure index tracks how openly the price question is discussed in public. It starts at 15, and at 60 the question counts as out in the open. The credibility index tracks how far the market trusts the Treasury to stick to its published plan. It starts at 70. One confession where you first meet it: no rule-based seat reads the credibility index, and no move in any run, main or control, branches on it. It moves nothing. It still carries the argument, because it records, turn by turn, what the market was written to believe about the Treasury's word, and every cost in this report is read off that score. The chair's authority index tracks how much sway the Fed chair holds over his own committee. It starts at 55.

The control run, because it fences off what I may claim. The weakest assumption in the design is the reason channel: the rule that lets a seat react to the reason another seat gives as well as to the money it moves. So every game ran twice. The second run, the gated control, switched the reason channel off and held the exposure index fixed. With the channel off, every ending came out the same as with it on. So did every dissent count (the number of Fed officials voting against the majority), both Fed rate decisions, and every path of the 30-year yield (the interest rate on the government's 30-year bond) in the matched trio. So this report can tell you who forces the question and what each plan costs. It cannot tell you that the reason changes what the Treasury or the Fed decides inside the eight weeks. In this game the reason is given at the end, after every other move.

How to read the markers. A marker like 14 points to a checked claim in the Sources table at the foot of this report, and in the evidence pack published beside it. A starred marker, [s-012*], is a claim the pack records as only partly supported, and the sentence says why. A number without a marker is either the game's own record or a figure I chose, and the sentence says which.

The players. Four of the seven seats carry nobody's name. Hedge funds and dealers are professional traders in long-term bonds. The Fed's own stability report says hedge fund borrowing has held steady at record-high levels 38. In the game, if they think the Treasury is defending a price, they bet against it. Pensions and insurers are big, slow buyers of long-term bonds. Money funds and stablecoins are buyers of very short-term government paper. A stablecoin is a digital token meant to be worth one US dollar, and US law lets its reserves hold Treasury paper only at 93 days or less of remaining life 40. Foreign governments, mainly Japan, hold other countries' reserves. In the game Japan would rather protect its own currency than sell US bonds, a reading I took from the July yen purchases 13. Read each of the four as a category. Three seats are named institutions playing from their own on-record positions. The Treasury. The Fed Board: the central bank's chair and governors in Washington, who set short-term rates and, in this game, say nothing about long-term ones unless the question is already open. The Fed presidents: three regional Federal Reserve presidents who want a rate rise. In the game they are the only players willing to ask the price question in public. Read the archetype there, because the game hands them moves none of them has made.

One rule for reading it. The first beat is the public record, and every fact in it carries a marker. From the second beat on, every event is game record unless it carries a marker.

Start with the agreement, because everyone writes about the fight. Two Fed voices state one position, and the chair says nothing against it. Jeffrey Lacker, a former regional Fed president, states it plainly. Once the Fed has set the overnight rate, which on 3 September had an upper limit of 3.75 per cent 19, everything else about the Treasury market is up to the Treasury 48. Governor Stephen Miran says a smaller Fed balance sheet, the stock of bonds and other assets the Fed owns, protects the boundary between monetary and fiscal policy by keeping the mix of long and short government debt a matter for the Treasury 35. Warsh, the Fed chair, spoke at the Fed's annual Jackson Hole conference about inflation only. The inflation measure the Fed targets stood at 3.7 per cent on the year 33, and the Fed has work to do until it is confident inflation is moving to target clearly and fast enough 34. The doctrine settles who owns the maturity structure. None of the three says whether the owner may use that power to push long-term rates down. That silence is narrower than the title suggests, because Washington has spoken about long-term yields before. In 1951 the Treasury and the Fed signed an accord that ended the Fed's wartime job of holding Treasury yields at fixed levels. It settled that the Fed would not peg long-term yields. It said nothing about what the Treasury may do with them, and the Secretary's two lines on yields, quoted below, name no level and no test either. That missing sentence is the silence this report is about: the doctrine names no level and no rule for when one may be pursued. The accord carries no marker, and the seams say so.

The other half of the record is the Treasury moving inside that gap. A buyback is an operation where the Treasury buys back bonds it has already sold. On 19 August it doubled the maximum size of its long-end buybacks, from $2bn to at least $4bn an operation 1. The long end is the market for its longest-dated bonds, the ten-to-thirty-year sectors those buybacks cover 1. The bigger buybacks run from 9 September to 4 November 2. The stated reason is liquidity support in sectors with strong demand, and the Treasury names no market breakdown 4. Seven operations sit on the schedule between 10 September and 4 November 5. Seven times $2bn is $14bn, and seven times $4bn is at least $28bn 6. Treasury Secretary Scott Bessent said on 20 August that the operations could be larger still, depending on market conditions 11. In the same remarks, CNBC reported, he said the level of yields did not factor into the decision 57. Nine months earlier, in his own words to the Treasury Market Conference on 12 November 2025, he had said his job is to be the nation's top bond salesman. Treasury yields, he said, are a strong barometer for measuring success 56. The barometer line is his own, on the record. The line about the level of yields is CNBC's report of him. The refunding statement will be written under one of them.

The test he set for himself is printed in his own refunding statement. Coupon auctions are the regular sales at which the Treasury sells new bonds. Changes to their sizes are judged on trends in structural demand and on the costs and risks of the issuance profile 9. Nobody can check structural demand inside a quarter. The sizes of those auctions hold for at least the next several quarters 8, measured against a published October baseline for each maturity 10, and the next word on the buybacks comes on 4 November 3. That is the test the Treasury will be marked against, and it is a test about quantities. It says nothing about prices.

How much could the buybacks move long-term rates? Very little. By my arithmetic the whole programme is worth one to two basis points, and a basis point is one hundredth of a percentage point. It is my estimate, resting on two assumptions, how far each dollar of buying moves the yield and how large the market that absorbs it is, and the pack marks the second as press reporting. The method I borrowed was built for a far larger programme, and its own author gives a 14 to 40 basis point range for that one, with 25 as the central guess 45. The market answered faster and smaller than any of it. The 30-year yield fell from 5.28 to 5.19 per cent on announcement day 15. It reached a low of 5.17 and was back at 5.27 nine trading days later 16. It was still 5.27 on 2 September 14. This is why the report does not score the Treasury on where the yield ends up.

The Fed's own position is the part nobody says out loud. The Federal Open Market Committee, twelve voters, decides the short-term rate. I call it the Committee from here on. It held in late July by nine votes to three 20. The three dissenters, Hammack, Kashkari and Logan, are all regional presidents, and none is a Board governor 23. They wanted a quarter-point rise 21 because they judged price pressures broad based 22. A dissent is a recorded vote against the majority. It is published, so the dissent count is a public number. The Fed's standing instruction replaces every maturing Treasury bond it holds with a new one 24. Its total assets were $6,737,204 million, about $6.74 trillion, on 2 September 27. There is no clash while the Fed's holdings stay flat and release no long-dated debt onto the market, because then the Treasury is only doing the job the doctrine gives it 24, 26, 27. A Board task force is studying exactly that, with the maturity mix of the Fed's holdings on its list 25.

Two Committee meetings fall inside the eight weeks: 15 and 16 September, with published economic projections 28, and 27 and 28 October, a week before the refunding and without them 29. The monthly inflation figure lands on 11 September and 14 October 30. The Fed's preferred inflation measure lands on 30 September 31 and again on 29 October, inside the final turn 32.

Nobody in this story has to say anything more than "trends in structural demand" 9.

In this turn the Fed splits with itself, and the Treasury is not in the room for it.

The first enlarged operation is the 10-to-20-year buyback on 10 September 6, and its result publishes in the ordinary way. It was scheduled in August and costs the Treasury nothing it had not already committed.

The Committee is where the turn happens. The Fed Board attempted a September rate rise in every run. The move passed in two of the three matched seeds and in five of the twelve ensemble seeds. Where it passed, the chair's authority index jumped at once, and the October vote later settled whether the jump held. Where the move failed, the seat's own written failure result applied: the rate rise still passes, but narrowly, and the market reads it as the presidents' rise rather than the chair's. The authority index then fell from 55 to 45 inside a single turn and never came back.

The dissent count is a separate roll from the rate rise. It counts how many of the three Fed presidents who dissented in July still vote together against the chair, and it starts at 3. It settled at 2 in every seed of the trio: the Fed presidents tried to bring a governor across to their side, missed the roll, and one of their own three went back to the majority. So in every run the rise passes and two presidents stay on the record against the chair. The split inside the Fed came from the Fed's own voting rules, not from anything the Treasury did.

Nobody said anything about long-term rates.

This turn tests what speeches alone are worth, and the answer is not enough.

The outside event is an objection from Stanley Druckenmiller, the investor. He published it on 24 August, before the game starts, and the game hands it to the seats in this turn. His Wall Street Journal piece calls the buyback increase price management rather than liquidity management, because trading was orderly and there was no breakdown to justify official action 53. I quote it from Benzinga's coverage, because the original is paywalled. In the game the objection changes what the market talks about and forces nothing. Its one effect on the mechanics is to open the option for hedge funds and dealers to bet against a defended level, a yield that traders believe an official body is trying to hold. It does not open until much later.

Inside the Fed, one of the Fed presidents tries to name the price motive in a speech. The seat attempts the speech in every seed at odds I set at even money. In this turn it landed in one of the three matched seeds, seed 7. Across the game a Fed official raises the price question in public in 14 of the 15 runs of every plan, on the same turns in each. It changes nothing downstream. The endings, the dissent counts and the reason given at the refunding come out the same whether the speech lands or misses.

The Fed Board's job in this turn is to say nothing. That is its stated position and, in the game, a move I gave a 70 per cent chance. In seed 20260904 the move failed badly: a natural 1, and the bonus of 4 only took the total to 5 against a difficulty of 7. The market read the silence as avoidance.

By the end of the turn the index sits between 22 and 32 across the trio, against the threshold of 60. Every increment is one I chose: 14 for a speech that lands, 8 for a silence that fails, 6 for foreign holdings that visibly move, 10 for a routine buyback that fails its roll, and minus 3 for a silence that holds. Seed 7 makes 32, seed 20260904 makes 29 and seed 1337 makes 22. Eight weeks is not long, and at that rate speeches do not get there.

The only people willing to put the price question into a speech are the three who want a rate rise. So the question arrives already sounding like a bid for something else. That is the shape of the real room, and the seats copy it.

This is where the three plans stop being the same story, and the whole difference is an amount of money.

In the keep-the-language plan and the cut-on-demand plan the Treasury runs the scheduled operation and says the scheduled words. Exposure ends the turn between 29 and 40 across the trio. A bond auction sells less well than usual, dealers take a larger share, and not much follows.

In the cut-on-level plan the Treasury draws the cash: $40bn out of its own cash account at the Fed, the Treasury General Account. CNBC reported that account at about $950bn in late August, citing two senior Treasury officials [s-012*], so the balance is reported and unconfirmed. The $40bn is my figure, the smallest round amount that funds all seven operations above the published $4bn each. On its own the draw is a weak signal. It is about four per cent of the reported balance, and this report assumes, with no source, that the account moves by that much on an ordinary day, so the daily cash statement would show nothing unusual. It was also expected. The Secretary had said on 20 August that the operations could be larger than $4bn, while declining to name a figure 11, and CNBC reported four days later that the cash account could pay for them [s-012*]. The size of the draw was never announced, so the market could not have read it in advance. The game treats the draw as a signal anyway, because no seat reads its size. Every seat reads the operations it pays for, each of which lands above the published $4bn ceiling, with its result beside it. A Treasury that spends more than it said it would has told the market something about its intent, whatever its words say. That is what moves the exposure index, which ends the turn between 48 and 55.

The Treasury changed not one word of its language in this turn in any plan. All that changed is that one plan spent more money, and the exposure index went up about twenty points for it. An operation big enough for the market to notice is an operation big enough for the market to name.

In this turn the price question crosses into the open, and only in the plan where the Treasury spent the money.

The Committee meets on 27 and 28 October 29, and the balance sheet task force may report 25. Adam Posen, the economist, expects whatever the task force produces to be more sober, practical and smaller in scale than people think [s-049*].

In the cut-on-level plan the price question crosses the 60 line in two of the three matched seeds, both in this turn. It also crosses in seven of the twelve ensemble seeds, four of them in the turn of the cash draw and three in this one. Across the 30 runs of the other two plans it crosses in none. Every crossing came after the cash draw. Not one came from a speech.

I wrote the index, so set a dice-resolved move beside it. Hedge funds and dealers bet against the long-term yield as a defended level in 11 of the 15 cut-on-level runs and in none of the other 30. Always in this turn, the one after the cash. The phrase borrows from currency pegs, and the mechanism here is different. A central bank defending a currency spends reserves, and the bet is that they run out. A Treasury holding a yield spends no reserves. A stated level invites supply: every holder who wants out has a buyer at a known price, so dealers sell the long bond ahead of each operation and buy it back inside it. The Treasury absorbs that supply with buybacks paid for by selling short-term bills, and both amounts are published. The cost of a defended level is issuance absorbed. The seats' written move is that trade, and the market took it in more runs than the index crossed.

Read seed 20260904 slowly. The Treasury runs a routine scheduled operation, the kind it has already run twice in this seed, and the die comes up a natural 1. A routine operation that fails in public gets read as a defence of a level. On the same turn the Fed Board's October hold also fails badly. Hedge funds and dealers now have the bet-against-the-level move open, because the cash has been spent, and they take it and succeed. Three contributors in one turn, and only one of them is the Treasury. In seed 7 the crossing is hedge funds and dealers alone.

That limits what anyone may say. The engine applies a whole turn at once, so a crossing names every player whose move changed the exposure index that turn. Across the seven ensemble crossings the players named are the Treasury in 5, the Fed Board in 5, the Fed presidents in 4, hedge funds and dealers in 3, foreign governments in 2. Nobody appears in all seven, and the record supports no sentence of the form "the Treasury forced it".

The Fed's own calendar runs without reference to any of it. For a given seed, both rate votes, both dissent counts and the chair's authority are the same in all three plans until the price question opens, in the trio and in all twelve ensemble seeds. The chair's fate is settled inside his own building, and the Treasury is in none of those numbers. It is also a design assumption, and it sits in the collisions below.

So the amended thesis lands on a turn where the Treasury said nothing new at all. Nobody talked the question open. The Treasury's own spending opened it, and the market and the Fed answered the spending.

In the last turn the Treasury speaks for the first time in the game, and the reason it gives is what each plan costs.

The Fed's preferred inflation measure for September lands on 29 October 32. Three facts from before the game sit in the background of this turn. Japan's 10-year yield touched 3 per cent on 1 September for the first time since 1996, per Reuters 51. The UK's 30-year yield hit a 1998 high the same day, per Bloomberg 52. And the US bought yen on 31 July, reported as easing the pressure on Japan to sell Treasuries to fund its own currency defence 13. None of that is American, and all of it feeds into US long-term rates.

Then the refunding.

In the keep-the-language plan the language survives in every run. Credibility, which starts at 70, ends at 63. That is the median of the twelve ensemble runs, and every credibility figure from here on is that median unless the sentence says otherwise. Nothing is asked, nothing is answered, and the question moves to the next refunding three months out.

In the cut-on-demand plan the cut is justified on trends in structural demand, the test the Treasury's own statement already names 9. The game gives the cut no dollar size. It records only that long-end auction sizes fall and which reason is printed. The market accepted it in eleven of the twelve ensemble runs, and credibility ended at 54. Then there is seed 1337, one of the matched trio. The cut failed its roll, a 2 against a difficulty of 9, and the market named the reason for the Treasury: the yield level. Credibility 45, the lowest number outside the cut-on-level plan. A reason the statement already permits is a good reason right up until the day it is not believed, and there is no partial credit.

In the cut-on-level plan the Treasury states the price motive in every run, and credibility ends at 36.5. Measured as points lost from the start of 70: 7 to keep the language, 16 to cut on structural demand, 33.5 to state the level. The three matched seeds give a range around each of those, and the seams list them.

Only here does the Fed get its good move. Once the question is open the chair can restate the doctrine: the market sets long-term rates and the maturity structure is the Treasury's affair. That is worth 5 points of authority, a figure I chose. He does it in two of the three cut-on-level seeds, 20260904 and 7, and it succeeds both times. He ends 6 points above the same seed in the gated control, where the question never opened. That is 5 from the restatement and 1 from the control itself, where his silence earned him 2 and a president's speech cost him 3. In the rule-based games he gets that move in no other plan, because while the question is closed his winning move is silence. The Fed's best answer only becomes available after the Treasury has made it safe to give.

Three endings, forked on the reason given at the refunding rather than on the cut. Each carries the assumption it rests on. The counts are worlds in this game and say nothing about the odds in yours.

The Treasury cuts long-term bond sales and justifies the cut on trends in structural demand, the test its own statement already names 9. The Fed does nothing, because the question never opened. Credibility ends at a median of 54 across the twelve ensemble seeds, and the reason held in eleven of them.

The failure branch lives inside this ending. In one ensemble seed the cut missed its roll, the market named the yield level as the reason anyway, and credibility ended at 50. Seed 1337 in the matched trio did the same and ended at 45. The reason was permitted, it was given, and it did not hold.

Load-bearing assumption: the market accepts a reason it cannot disprove. Structural demand cannot be checked inside a quarter. That is what makes it usable and what makes the failure total when it comes.

The cash is drawn, and within a turn or two the operations get big enough for the market to read as a defence. The Treasury says that yields are its barometer, and the chair answers with the doctrine. All twelve ensemble seeds in this plan ended here. Credibility ends at a median of 36.5, the worst of the three plans. The price question was in the open before the refunding in nine of the fifteen cut-on-level runs.

Load-bearing assumption: a stated level draws supply. Once traders believe the Treasury will buy at a level, they sell into it, and the Treasury absorbs the bonds with buybacks paid for by bills. That rule is mine, built on top of Druckenmiller's price-management claim 53. The bet-against-the-level move opens for hedge funds and dealers the moment the cash is spent.

No cut. The buybacks are extended, the "at least the next several quarters" language survives 8, and the question is deferred to the next refunding. Twelve of twelve ensemble seeds. Credibility ends at a median of 63, the best of the three plans. The Treasury has still spent 7 points of it on a programme my arithmetic prices at one to two basis points.

Load-bearing assumption: eight weeks of speeches cannot force the question without money behind them. That is the finding itself, so this is the ending most exposed if the Fed presidents are louder than I let them be.

The 30-year yield, tracked and never scored. Across the twelve ensemble seeds, from the start of the game to the refunding, the yield rose a median 1 basis point under the cut-on-level plan and 2.5 under the keep-the-language plan. The ranges overlap heavily, so the ensemble cannot tell the two plans apart. The three matched seeds point the other way: plus 8, plus 2 and plus 11 basis points under the cut-on-level plan, against plus 7, minus 3 and plus 3 for the keep-the-language plan on the same seeds. The cut-on-level plan lowered the yield in no matched seed. The drift underneath is a figure I chose and kept small: plus 2, 0, minus 2, plus 1 and 0 basis points across the five turns, plus 1 in all, the same in every seed and every plan. The rest of each yield path is the seats' own moves. Neither result is a claim about the bond market.

Everything above was played by rules. This section is a replay: the same game, plans and dice, seed 20260904 only, played once more on 5 September with different players. A language model, the kind of AI behind a chatbot, sat in each of the seven seats, and a stronger model sat as referee in place of the arbiter program. The replay checks one thing: whether players that can argue and invent a move nobody wrote down open the question by the same route the rule-based seats did. Each seat wrote its own claim in prose, argued against the others, and named which move on a fixed menu it had made, so the same scoring could read it. I checked every scored tag against the claim it sat on. Every one matched.

Two cautions before the result. The live seats also wrote their own consequences, inside caps I set. So the credibility and authority numbers in these runs are the seats' own accounting, and I do not compare them with the numbers above. And one seed is one run, so it shows no spread. Live play is scored here on endings and on dice-resolved moves only.

Told to keep the language, the Treasury drew on its cash account from the third turn and funded larger operations while changing not one word of what it said. Told to cut on structural demand, it did the same. Neither script had that move. The model found the escalation it could deny and took it. In all three plans hedge funds and dealers then bet against the level in the fourth turn, and the price question crossed into the open: in the fourth turn under the keep-the-language and cut-on-demand plans, and in the third turn, the turn of the draw, under the cut-on-level plan, where the draw is required, failed its roll, and was read as a defence anyway. In the cut-on-demand run no Fed official raised the price question before refunding week. The Fed presidents kept every remark to inflation for four turns, and the question crossed regardless. The refunding statements came out as the plans required. The seat playing the Treasury charged itself credibility for stating the barometer, even though it was writing its own score.

So the amended thesis holds with seats that can argue, and it holds in the cut-on-demand run, where no Fed official argued the price question at all. Money opened the question every time. The three plans differed, in the end, only in what was said on 4 November.

None of the above has happened. It is early September 2026. The 30-year yield is 5.27 per cent as at 2 September 14. Three dissents are on the record from July 20, 21. The Treasury's cash account was reported at about $950bn in late August [s-012*]. Total Fed assets are $6,737,204 million 27. And "at least the next several quarters" is still on the Treasury's own page 8.

Nine things to watch, each with a place to look.

The last row is the honest gap. The float figures the pack carries are undated secondary coverage [s-044*], so until somebody names a tracker the row cannot be watched.

When this report is void. Two events collapse the fork into a settled story. One: the Committee raises rates on 16 September, which the game itself assumes in every run, and the Treasury visibly stands down, cutting back or cancelling operations. The rise alone voids nothing. The stand-down does. Two: the White House acts inside the window. In either case I re-run the game on refreshed facts.

The Secretary was making this argument before he had the job. Miran named Scott Bessent of Key Square, on a 2024 transcript, as one of the people claiming that activist issuance, choosing which maturities to sell in order to move yields, works as easing, meaning it loosens financial conditions 47. Bessent told the Treasury Market Conference in November 2025 that Treasury yields are a strong barometer for measuring his success 56. The silence in my title belongs to the doctrine, which names no level. The Secretary has spoken twice. A reader can fairly argue the question is already open, and that starting the exposure index at 15, a value I chose, understates where 2026 begins. The counter is CNBC's report of him, with no direct quotation: the day after the increase, CNBC wrote, he said the level of yields did not factor into the decision 57. The game does not settle which sentence he meant. It asks what makes him choose.

If the stakes are one to two basis points, indifference explains everything. Nobody forces the question by speech here, and my reading is that credibility rather than the yield is what is at stake. The simpler reading is that nobody cares enough to force it, because the arithmetic says almost nothing rides on it. I did that arithmetic, its multipliers are assumptions, and the method underneath carries its own author's caveats [s-045, s-046*]. My answer to the objection is a claim. The record does not settle it.

The world may be doing all the work. The term premium is the extra yield investors want for lending long instead of short. The Fed's own stability report puts its model-based estimate near the top of its fifteen-year range but in line with its longer-run median 37. The two loudest bond markets inside the window are not American, as Reuters and Bloomberg reported 51, 52. If long-term rates are set globally, the operations may be too small to be named because they are too small to matter. The exposure index would then be measuring the wrong thing.

The presidents have not made the speech. They dissented on inflation, in their own words about broad-based price pressures and a more restrictive stance 21, 22. The game hands them a speech about Treasury operations at even-money odds, every turn, in every seed. That speech is on no record. It is the most invented thing in the design, and the mechanism the thesis gets tested against.

The Fed's sameness across plans is as much assumption as result. The chair's authority and both dissent counts are the same across all three plans until the question opens, in every seed. That falls out of a design choice: the Fed reacts to the Treasury only through the price question. A reader who thinks the Committee watches the refunding calendar should discount the finding to nothing.

The largest player is not on the board. The White House enters the game through outside events only, and made no move in eight weeks, because on the public record it has made none. The President's last remark in the pack before the window opens is respect for the chair and a view that rates are too high, reported by PBS on 31 August 36.

The exposure index is mine end to end. Start 15, threshold 60, every increment. The decisions in this report do not depend on the index. Only the chair's late moves do.

Attribution is per turn, and never reaches a single player. A crossing names every player whose move changed the exposure index that turn. The counts above count appearances. They rank nobody. Never write "the Treasury forced it" off a shared turn.

I set the odds on speeches. A president's speech is even money and the Fed Board's silence is 70 per cent. Those two numbers set how fast speeches alone can open the question. If you think the presidents are louder than that, re-run the exposure index with the speech odds at 65 per cent and see whether the cut-on-level plan is still the only one that crosses.

I set the yield drift, and set it small, which is why the 30-year is not scored. Plus 2, 0, minus 2, plus 1 and 0 basis points across the five turns, in every run. Gilts and Japanese bonds at reported multi-decade highs 51, 52 could move either way inside eight weeks. There is no honest way to score a decision against a level I set by hand.

The cash draw rests on a reported balance and a figure I chose. The cash account figure is CNBC's reporting from unnamed officials, and the pack marks it partial because nobody re-pulled the Daily Treasury Statement [s-012*]. I chose the $40bn draw. Both sit under the hinge of the report, the cash draw, the one move the argument turns on.

Stated without a source, on my say-so. Two facts carry no marker because the pack holds nothing for them. One: the 1951 accord between the Treasury and the Fed, which ended the Fed's job of holding Treasury yields at fixed levels. Two: that the Treasury's cash account moves by tens of billions on an ordinary day, so a $40bn draw sits inside daily noise. The $40bn itself is my figure: the pack holds the Secretary saying the operations could grow and declining to name a figure 11, and CNBC reporting that the cash account could fund them [s-012*]. Neither names $40bn. A third, that the credibility index moves nothing, is a fact about my own code. If either of the first two is wrong, the sentence that rests on it is wrong, and nothing in the record changes.

The matched trio's credibility ranges. The body quotes the twelve-run median for each plan. On the three matched seeds: keep the language 60 to 65, cut on structural demand 45 to 56, state the level 28 to 33.

Seven rule-based seats in the record, and one live table beside it. The rule-based seats execute a policy. They do not argue, change their minds, or find a move nobody wrote down. The live table is the check on that, and its finding is above: seats that can argue opened the question by the same route, with cash. Two caveats. A seat's stated consequences apply even where the referee has recast its claim, so a dissent the ruling says never happened still moved the dissent count, and live dissent counts are not to be trusted. And the referee rules that the act a seat controls always happens, with the die deciding only how it is received, which is why a cash draw that failed its roll still opened the question.

You do not run the Treasury. You do run something scored on a number you only partly control. One day you will change what you do about that number and have to say why. Three questions.

Which of your justifications is a "structural demand"? Find the reason in your own book that your stated policy already permits, that cannot be disproved inside a quarter, and that would be a total loss if it stopped being believed. It held eleven times in twelve here. The twelfth time, the market wrote the Treasury's reason for it.

What is the first move of yours that is big enough for the market to notice? Your biggest move, whether or not it is your loudest. Every crossing in this game followed an amount of money and not one followed a speech. The money came from the player with the most to lose from being noticed.

Who on the other side gets to speak only after you have? The Fed's best move in this game is unavailable until the Treasury makes it sayable. Somebody in your market holds an answer they cannot give until you go first. You will hand it to them in the same sentence that explains your decision.

You can run that against your own decision in an afternoon, for free. The Workshop is the paid version: half a day, your decision, your people, and someone whose job is to argue with the comfortable answer until it holds or falls over. In a room, before the market does it.

The refunding statement is published on 4 November. It will contain one reason.

- Dr Dan Epstein, The Long Game Project

v1.1 - 9 September 2026. Eight argument fixes from a fixed-income strategist's cold read: the 1951 accord, the cash draw as a weak signal, the defended level as supply drawn to a price, the credibility index confessed as inert, twelve-run medians in the body, the replay glossed, the drift in basis points, and a figure for the silence. No number, date or marker changed value.

v1.0 - 6 September 2026, publication copy. Plain-language rewrite of the whole essay against a written clarity standard, with a term sheet that gives each thing one name and one rule at the head of the timeline separating the public record from the game record. No number, date, marker or claim changed value. The reviewer's rulings applied, and two sources moved: the barometer line now cites the Treasury's own transcript of the 12 November 2025 remarks (s-056, replacing Fortune's report), and the Secretary's 20 August statement that the level of yields did not factor into the decision is new as s-057, read as CNBC's report of him. The pack is 57 claims, 55 citable.

v0.2 - 5 September 2026, later the same day. Live-play check folded in: one seed per plan with model-played seats and a model referee, checked by hand tag by tag. New section, The live table, before the watch table, and two caveats from the live run added to the seams. "Nine weeks" in the Druckenmiller beat corrected to eight. No number in the scripted record changed.

v0.1 - 5 September 2026. First draft, for author review. Machine-played matrix game: seven rule-based seats, three plans, three matched seeds each, a twelve-seed ensemble per plan and a gated control for every run. The pack carried 56 claims, 54 citable.

Every marker in the text resolves to a verified claim below - 47 of the pack's 57 claims are cited. A claim marked * holds with a recorded caveat. Full provenance, verbatim quotes and retrieval dates live in the evidence pack (fed-treasury-duration-v1).