The Account Printed a Record. Then the Price of Cash Moved.

The Treasury's general account closed at $1,015.6bn on 28 October, from $945.2bn a week earlier, having printed $1,052.5bn the day before - the highest close in the series' four and a half years. Reserve balances fell $68.8bn over the same seven days to $2,893.8bn, the lowest since December. The two differ by $1.6bn, which is what a transfer looks like. This week the price moved with it: secured overnight money fixed 3.94% against an administered 3.90%, positive for four sessions after seven weeks at or below zero, with $0.35bn left at the reverse repo facility (as of 29 Oct). Gold fell $38.20 on the session.

There are two prices of money in this system. One the state sets by announcement: the rate it pays on the balances banks keep at the Fed, currently 3.90%. The other the market sets by lending: the rate at which cash is swapped for Treasury collateral overnight, called SOFR. For seven weeks the second traded at or below the first, which is what a system with spare cash and no use for it looks like. This week that stopped. The Treasury drained $68.8bn of reserves into its own account, took that account to a record close, and overnight money fixed 4bp above the rate the state administers. Nothing was announced. The price simply found where the cash was scarce. That is the governance content: once the sovereign's cash management is large enough, it stops being a background transfer and starts setting a price. Ahead: the flow, the price, the long end that charged for it, and what would show this read is wrong.

The account printed a record close and held a trillion for three sessions

The Treasury's general account closed at $1,015.6bn on 28 October, from $945.2bn on 21 October. That is $70.4bn into the state's cash balance in seven days.¹

The peak came a day earlier: $1,052.5bn on 27 October, the highest close in the series, which begins in April 2022.¹ Closes above a trillion ran from 26 October through 28 October. The last time the account held three consecutive closes above a trillion was April, and that run also stopped at three.¹

Read the level before the move. An account above a trillion is roughly a third of the reserve base sitting in a chequebook. The Treasury does not hold that because it needs the cash on Tuesday. It holds it because building the balance is how the sovereign withdraws money from the banking system without legislating anything.

This is the point where cash management stops being plumbing and becomes policy. The decision was a debt-management choice, taken months ago, about how much cash to hold against the next fiscal cliff. The execution is a series of settlement dates nobody votes on. The effect lands on the reserve base, which is the number every other price in this market is marked against.

The reserve base paid, and the buffer is empty

Reserve balances were $2,893.8bn (as of 28 October), from $2,962.6bn on 21 October. That is $68.8bn out of the banking system in one week, and the lowest print since the week of 3 December 2025.²

The two numbers differ by $1.6bn. The cash did not leave; it changed hands, from the banking system to the sovereign. The reverse repo facility went from $0.24bn on 21 October to $0.35bn on 29 October, so it is not parked there either.³

That is what separates this week from the sequencing of the past two years. Reverse repo used to be the shock absorber: the Treasury could build cash and the money came out of the facility instead of the reserve base, and funding markets noticed nothing. There is no absorber left. Take-up is a rounding error against a $70bn call, so the call lands on reserves dollar for dollar.

A system with an empty buffer has no slack between a debt-management decision and the price of overnight money. Hold that as a picture before the price numbers arrive.

After seven weeks at or below zero, the price crossed

Secured overnight money fixed at 3.94% on 28 October against an administered rate of 3.90%. That is +4bp.⁴

It had not printed a positive fixing since 3 September. Between 4 September and 22 October, 33 consecutive fixings came in at or below the administered rate - 27 negative and six at zero - and the last of them, 22 October, closed at exactly the administered rate rather than below it.⁴

The run then went +2bp on 23 October, +3bp on 26 October, +5bp on 27 October and +4bp on 28 October. That is four consecutive positive fixings - only the second run of four or more this year, after fourteen sessions from 13 February to 5 March. The last fixing at or above 5bp before this week was 15 April, at +7bp.⁴

The unsecured market did not follow. The effective fed funds rate is 3.88% and has not moved for eight consecutive sessions back to 19 October. Secured money now sits 6bp over unsecured - the ordering that says collateral, not cash, is carrying the price.⁴

Volume rose with the rate, from $2,942bn on 21 October to $3,074bn on 28 October.⁴ A rate that rises on rising volume is demand. A rate that rises on thinning volume is a quote nobody traded against. This is the first kind, and the size is modest: four basis points is a scarcity signal, not a funding squeeze.

Last Friday's version of this trade has expired

A week ago this column closed on the observation that the largest cash call of the sequence had been absorbed without a scarcity premium, and that the absence of a premium was itself the governance content: the sovereign could move $59bn by announcement and the price of cash would not move.

That observation is now falsified on its own terms. The same transfer ran larger - $70.4bn instead of $59.3bn - and the price moved, by four basis points, in the same week. What changed is not the size of the call but the stock of spare cash available to answer it. The buffer was $1.2bn on 15 October; it is $0.35bn now.

This is the useful kind of correction, because it is testable rather than rhetorical. The prediction embedded in last week's read was that the transfer could repeat indefinitely without a price consequence. It repeated once and produced one. The mechanism was not wrong; the buffer assumption underneath it was.

The long end charged the 20yr more than the 30yr

The curve bear-steepened over the week. The 2yr fell 4bp to 4.61%, the 10yr rose 2bp to 4.98%, the 30yr rose 3bp to 5.32%, and the 2s10s slope went from 30.3bp to 37.9bp on the site's own series.⁵

The 20yr is at 5.35% against the 30yr at 5.32%. On the par curve the 20yr paid 4bp over the 30yr on 28 October, from 5bp on 21 October - still inverted, modestly narrower.⁶

An inversion inside the long end is not a recession signal. It says the marginal buyer wants more compensation for the 20yr point than for the 30yr point, which is where auction sizes concentrate and where the pension bid is thinnest. The sovereign chooses the maturity mix; the market prices the consequence.

Put the two ends together and the read is coherent. At the front, the state withdrew cash until secured money paid a premium. At the back, the state issues into a buyer base that charges more for the point it issues most. Same actor, same direction, and both prices moved this week.

The structure ranked the charter above the coin

STRC closed at 99.99 on 29 October, +0.07 on the session and +1.34 on the week - percentile rank 100.0, the top of its own 316-session history.⁷

MSTR closed at 141.62, -3.54 on the session and -4.18 on the week, rank 23.1. STRK at 70.22 ranks 32.8; STRF at 102.67 ranks 75.1.⁷

A preferred printing at the very top of its range while the common sits in the bottom quarter is not a bull signal on the underlying. It is a ranking. Holders are paying up for the contractual claim and declining the residual, and the gap between those two answers is the market's estimate of how much of the structure's value sits in the charter rather than the coin.

Bitcoin is $78,477.74, down $2,789.84 on the session, with a 30-day trend of -6.14%.⁸ The instrument that owns the coin fell; the instrument that owns a claim on the issuer rose to a record. Same issuer, opposite verdicts, in a week when cash got a price.

Where this lands on the tape

The composite headline is +0.304, regime NEUTRAL, coverage 5/5 sub-indices - down from +0.357 a week ago, and back inside the neutral band after sitting in tightening.⁹

Rates & Policy at +0.919 is dominant. Liquidity is -0.221 from -0.430, Credit & Stress +0.019 from +0.449, Growth & Inflation -0.052, Scarcity +0.855.⁹

Volatility priced none of it. The MOVE is 93.54, down 13.30 on the week, 45th percentile; the VIX is 17.20, down 1.83 on the week, 13th percentile.¹⁰

Equities were higher: the S&P at 7,720.39, up 56.96 on the week in the 97th percentile, the Nasdaq 100 at 29,402.51.¹⁰ Credit spreads tightened: high yield to 3.14% on 28 October from 3.27% on 21 October, investment grade to 0.86% from 0.88%.¹¹

Gold is $3,993.20, down $38.20 on the session and $44.20 below the 23 October print of $4,037.40.⁸ The asset that is nobody's liability fell in the week the sovereign's own cash balance set a record. That ordering is consistent with everything above it.

Nothing in the tape is stressed. That is the point worth registering. A 4bp scarcity premium at the front and a persistent long-end inversion are not events; they are a repricing of who pays for the state's cash. Volatility stays low because the move is orderly, which is how a structural change in the price of money usually arrives.

Where this breaks

The price read dies if SOFR returns to or below 3.90% and holds there for a week with the reserve base still near $2,893.8bn. That would make four basis points a settlement-date artefact, and the par prints of 21 and 22 October the true reading.

The flow read dies if the next reserve print is back above $2,962.6bn while the account still holds near a trillion. Reserves cannot fall from a transfer that did not happen, so a reserve recovery against a full account would mean the drain came from somewhere else - currency in circulation, the foreign repo pool, or a Treasury balance held outside the general account.

The long-end read dies if the 20yr goes back under the 30yr with both above 5.40%. That would make the inversion an auction concession rather than a buyer base, and the par prints of 5.41% and 5.37% on 28 October are the levels to test.

The structure read dies if STRC falls back below 99.92 while the common holds above $145.16. Preferred down and common up is a risk-on rotation inside the structure, and it would make this week's record a flow rather than a ranking.

The confirming case is SOFR above 3.90% for a second week with reserves under $2,893.8bn, the account near a trillion and the 20yr still over the 30yr. Three prices, one actor.

Watch the reverse repo line above all. Take-up back above $50bn restores a bid under the floor and removes the mechanism; take-up near zero with the account at a trillion means the next call lands on reserves with nothing in between, and the premium widens from four basis points.

Notes

1. Treasury general account closes, Treasury Fiscal Data operating cash balance: $1,015.6bn (28 Oct), $1,052.5bn (27 Oct), $1,020.7bn (26 Oct), $982.5bn (23 Oct), $945.2bn (21 Oct). Series begins 18 April 2022, 1,136 closes; 27 October is the highest close by value. Prior three-close run above $1trn: 20, 21, 22 April 2026.

2. Reserve balances $2,893.8bn (FRED WRESBAL, week ended 28 Oct), from $2,962.6bn (week ended 21 Oct), $68.8bn. The most recent prior weekly print below $2,893.8bn was 3 December 2025 at $2,858.3bn.

3. Reverse repo take-up $0.353bn (FRED RRPONTSYD, 29 Oct) against $0.237bn (21 Oct). Residual between the account build and the reserve drain: $1.6bn.

4. SOFR 3.94% (28 Oct) against IORB 3.90%, +4bp; fixings 22 Oct 3.90, 23 Oct 3.92, 26 Oct 3.93, 27 Oct 3.95, 28 Oct 3.94 (NY Fed reference rates; FRED IORB). Volume $3,074bn (28 Oct) against $2,942bn (21 Oct). EFFR 3.88% unchanged from 19 to 28 October. Between 4 September and 22 October, 33 fixings carried no positive spread; the last positive fixing before 23 October was 3 September at +1bp. Prior fixing at or above +5bp: 15 April at +7bp.

5. Site series, 29 October: 2yr 4.61% (-0.06 on the session, -0.04 on the week, percentile 74.1); 5yr 4.78% (+0.00, 81.5); 7yr 4.87% (+0.01, 81.0); 10yr 4.98% (+0.02, 80.4); 20yr 5.35% (+0.03, 78.8); 30yr 5.32% (+0.03, 79.8); 2s10s 37.9bp from 30.3bp. Real 10yr 2.639% on a 2.35% breakeven (FRED T10YIE, 29 Oct).

6. Treasury par curve, 28 October close: 3M 4.18%, 2yr 4.67%, 5yr 4.85%, 7yr 4.95%, 10yr 5.05%, 20yr 5.41%, 30yr 5.37%; 2s10s 38.0bp on the rounded par prints. The 20yr over the 30yr was 4bp on 28 October against 5bp on 21 October (5.47% and 5.42%).

7. STRC 99.99 (+0.07, +1.34 week, percentile 100.0 of 316); STRF 102.67 (-0.10, percentile 75.1 of 402); STRK 70.22 (-0.25, percentile 32.8 of 438); MSTR 141.62 (-4.18, percentile 23.1 of 753). All 29 October.

8. Gold $3,993.20 (-38.20 session, percentile 61.2) against $4,037.40 on 23 October, $44.20 lower on the week. Silver $51.365 (-2.346 week); copper $6.5405 (+0.0935); WTI $102.96 (+0.96); Brent $112.38 (-2.50 session, -15.92 week); DXY 101.857 (-0.183). BTC $78,477.74 (-2,789.84 session, -6.14% 30d, -2.00% 5d, 48.7); ETH $2,249.96 (-107.66, 22.8).

9. Composite headline +0.304, regime NEUTRAL, dominant Rates & Policy, coverage 5/5 sub-indices. Sub-indices: Rates & Policy +0.919, Liquidity -0.221, Credit & Stress +0.019, Growth & Inflation -0.052, Scarcity +0.855. Previous week, 23 October: headline +0.357, regime TIGHTENING, Liquidity -0.430, Credit & Stress +0.449.

10. S&P 500 7,720.39 (+56.96 week, percentile 97.3); Nasdaq 100 29,402.51 (+443.82); QQQ 715.34 (+10.87); VIX 17.20 (+0.29 session, -1.83 week, 13.2); MOVE 93.54 (-13.30 week, 44.9); TLT 81.49 (-0.25, 42.0); LQD 104.22 (-0.06); HYG 77.58 (+0.48); JNK 93.19 (+0.68).

11. High-yield OAS 3.14% (FRED BAMLH0A0HYM2, as of 28 Oct) from 3.27% (21 Oct); investment-grade OAS 0.86% (BAMLC0A0CM, as of 28 Oct) from 0.88%. Excess spread 2.28%. Initial claims 194k (FRED ICSA, week ended 24 Oct) against 195k (week ended 17 Oct); continuing claims 1,701k (week ended 17 Oct) against 1,722k (10 Oct); CPI 3.72% and core 2.57% year on year (September).

12. Both charts are generated by script from this piece's own series: the flow from the reserve and account prints of 21 and 28 October and the reverse repo take-up of 21 and 29 October; the spread path from the ten most recent SOFR fixings against the IORB of the same date. Neither is hand-drawn.