Reserves Built $36.3bn. The Chequebook Paid Out $79.4bn. $43.1bn Did Not Arrive.
Two balances explain most of the dollar system. Reserves are the cash banks keep on deposit at the Federal Reserve. The general account is the Treasury's chequebook at the same institution: money spent from it arrives in the banking system as reserves, money raised into it is pulled back out. Last Tuesday this column made a conditional call on that mechanism - the chequebook had just released $64.0bn, so the next reserve print should rise by tens of billions, with the arithmetic pointing to roughly $2,952.9bn as a ceiling. The print rose $36.3bn. Direction right, size wrong, and the $43.1bn gap is now the most informative number in the snapshot. Ahead: why a second consecutive miss changes the model from an estimate into a bound, why the parking lot that looked like a cushion last week was a month-end artefact, and what the collapse in the collateral premium says about who was actually short.
The build arrived. The size did not.
Reserve balances were $2,930.1bn as of 4 Nov, from $2,893.8bn as of 28 Oct. That is $36.3bn on the week.1
It is the first positive weekly print since 7 October. The four before it ran +81.6, -7.5, -59.5 and -68.8. A turn of $36.3bn after two months of drain is a turn.1
The level is $36.3bn above the 2026 low of $2,893.8bn recorded a week earlier, $32.5bn below the 21 Oct print of $2,962.6bn, and $99.6bn below the 7 Oct high of $3,029.7bn.1 One week has not undone four.
The chequebook paid out more than twice what the banks received
The Treasury's general account closed at $1,015.6bn on 28 Oct and $936.2bn on 4 Nov. That is $79.4bn out of the account across the 7 days ending on the reserve date.2
Money leaving that account lands in the banking system as reserves, so the payout is the model's only input. $2,893.8bn plus $79.4bn points to $2,973.2bn. The print was $2,930.1bn. The shortfall is $43.1bn.1,2
Last Tuesday the same subtraction left $31.2bn unexplained against a $64.0bn payout. This week leaves $43.1bn against $79.4bn. Two misses, one direction, and the second is the bigger share of the cheque: 54.3% against 48.8%. A residual that grows as a share of the move is not a rounding error. It is a second flow, and it is now the largest single unknown in the funding arithmetic.
The ceiling missed by $22.8bn too
Last week's note was explicit about the size. $2,893.8bn plus $64.0bn of chequebook payout, less $4.9bn absorbed by the parking lot, pointed to roughly $2,952.9bn, and the note called that a ceiling on the drain rather than a forecast of the level.1,2,3 The print was $2,930.1bn.
So the ceiling held as a ceiling, by $22.8bn, and that is the only part of the call that survived. The direction was right because a $79.4bn payment cannot be invisible. The magnitude was wrong for the same reason it was wrong last week, and the error is not shrinking.
That is worth stating without decoration. A one-for-one model that misses high one week and misses the same way the next is not converging on the level; it is telling you a liability the snapshot does not carry is doing steady work. The honest form of the claim is now: the chequebook sets the sign of the reserve move and only part of its size.
What the snapshot cannot see is the story
Two balances are visible and both are counted: reserves and the chequebook. Currency in circulation, the foreign repo pool and the non-reserve deposit lines at the Federal Reserve are not in this snapshot. The $43.1bn sits in one of them, and this column can name candidates but not the amount in each.1,2,3
Currency demand is the boring explanation and the likelier one for a week with a settlement calendar shifted by a holiday. A counterparty accumulating dollars outside the reserve system is the worse one. Both produce the same print, and the print alone cannot separate them.
What separates them is the price of overnight collateral, and that has already answered. If a real dollar shortage were building outside the reserve system, secured money would not have gone quiet on the same week reserves rose.
The cushion was a month-end artefact. This column got that wrong.
Overnight reverse repo take-up was $0.252bn on 9 Nov, from $0.150bn on 6 Nov, $0.125bn on 5 Nov, $0.225bn on 4 Nov and $0.427bn on 3 Nov.3
Last week this column read the $5.087bn print of 2 Nov as the facility waking up: a shock absorber worth 8% of the payout, and the one line in the snapshot cutting against the squeeze. It was $0.427bn two sessions later. The reading was wrong, and it belongs in the record rather than a footnote.3
What the print measured was the calendar. Take-up printed $11.539bn on 30 Sep, the quarter-end, and $5.087bn on 2 Nov, the first business day of the month. Both spikes lasted one session and both were gone by the next. Balance-sheet dates push cash into the facility; the session after pulls it out. At $0.252bn on 9 Nov it is functionally zero.3
So there is no buffer under the next build. Every dollar the chequebook takes out still comes out of reserves, and every dollar it pays out has to arrive without a cushion to absorb the timing.
The collateral premium ended at ten sessions
Secured overnight money fixed at 3.89% on 6 Nov against an administered rate of 3.90%, a spread of minus 1 basis point.4
That ends 10 consecutive positive prints: +2, +3, +5, +4, +4, +5, +5, +7, +4, +2, from 23 Oct to 5 Nov. Before the run the spread printed +0 on 22 Oct and -2 on 21 Oct.4
The squeeze this column tracked for two weeks is over, and it ended the way funding squeezes end: with the cash arriving. The run covered the $128.4bn two-week drain into the 28 Oct low and broke on the first week reserves actually rose. The price of overnight collateral is the cleanest confirmation the mechanism has produced, and it is now saying the opposite of what it said for ten sessions.1,4
Unsecured money never moved. Effective fed funds fixed at 3.88% on every session from 26 October to 6 Nov.4 On 3 Nov secured money traded 9 basis points above it; on 6 Nov it trades 1. Cash was never scarce. The paper it is lent against was, and that has stopped too.
The front end paid for the funding repair
The 2yr closed at 4.56% and the 3-month bill at 4.24%, with the 5yr 4.67%, the 7yr 4.76%, the 10yr 4.88%, the 20yr 5.29% and the 30yr 5.26%. All levels are the 9 Nov par close; the snapshot was taken at 07:00 UTC today.5
On the session the bill added 5.0 basis points and the 2yr 5.0, against the 5yr 3.0, the 7yr 2.0, the 10yr 2.0, and the 20yr and 30yr unchanged at 5.29% and 5.26%.5 A bill and a 2yr selling off is a market putting the overnight rate back up after a squeeze, which is what the funding data says happened.
On the week the order reverses. The bill added 8.0 basis points while the 2yr was unchanged at 4.56%, and the long end rallied: the 5yr -4.0, the 7yr -4.0, the 10yr -6.0, the 20yr -5.0 and the 30yr -5.0.5
2s10s is 32.0 basis points, from 35.0 on 6 Nov and 38.0 on 30 Oct.5 The corridor has round-tripped in 7 sessions: 38.0 on 30 October, 37.7 on the 5 November cash marks, 35.0 on the 6 November par close, 32.0 now.5,6
Every basis point of the weekly fall came out of inflation
The 10yr breakeven is 2.22% as of 9 Nov, from 2.23% on 6 Nov and 2.34% on 30 Oct.7 That is 12.0 basis points across 6 sessions, and 2.22% is the lowest print since 5 Aug, which also closed at 2.22%.7
The nominal 10yr fell 6.0 basis points over the same week. Split it: the breakeven contributed -12.0 and the real yield +6.0, to 2.660% from 2.600%.4,7 Inflation compensation supplied more than the whole move; the real yield gave a little back.
On the session the legs ran the other way, the breakeven -1.0 and the real +3.0. A real 10yr at 2.66% against a breakeven at 2.22% is a market pricing tight money and contained inflation at once. That is not a growth scare, and the equity tape agrees.5,8
Credit tightened off a 2026 wide and cash confirmed
High-yield OAS is 3.42% as of 6 Nov, from 3.48% on 5 Nov - the 2026 high - and 3.44% on 30 Oct. That is 6.0 basis points of tightening in one session.9
Investment-grade OAS is 0.93% as of 6 Nov, against a 2026 high of 0.94% on 2 Nov; it has not moved off the wide.9
Cash agreed. HYG closed at $77.51, up $0.28; JNK at $93.01, up $0.31; LQD at $104.67, up $0.14; TLT at $81.96, up $0.04.8
The labour data is not the reason for the tightening. Initial claims were 189k in the week to 31 Oct, continuing claims 1,718k in the week to 24 Oct, payrolls +165k in October with unemployment at 4.1%, and core CPI 2.57% year on year on a 3.72% headline for September.10
6.0 basis points off a 2026 wide against 189k claims is a funding story, not a default story. The default story would need claims to lead and spreads to follow; here the spread led and the claims never moved.
Risk took the whole move
The S&P closed at 7,549.94, up 173.08 on the session and +8.54 on the week, at the 92nd percentile of its range. The NDX at 27,907.63, up 917.64.8
Volatility collapsed into it. VIX is 19.92, from 25.64, at the 19.9th percentile; MOVE 85.98, from 92.77 and 14.38 lower on the week, at the 35.9th.8
Gold is $4,129.30, up $14.80 on the session and $191.10 on the week, +4.9%. Silver $53.33, copper $6.325. WTI $85.28, down $13.61 on the week; Brent $91.70. The dollar index 101.646, dollar-yen 158.00.8
An equity market at the 92nd percentile, gold up 5% on the week and a real 10yr at 2.66% are not three expressions of one view. Two of them hedge the third, and the composite cannot see that.
The composite went neutral on price, not quantity
The headline composite is +0.297, regime NEUTRAL, from +0.496 TIGHTENING. Sub-indices: Rates & Policy +0.662, Liquidity -0.563, Credit & Stress +0.457, Growth & Inflation -0.045, Scarcity +0.975 and dominant.11
Two blocks fell and one rose. Credit & Stress went +0.907 to +0.457, and it was almost entirely VIX: the VIX term went +1.885 to +0.582 as the index fell from 25.64 to 19.92.8,11
Liquidity went the other way, -0.147 to -0.563, because a looser funding market scores more negative on a scale where positive is stress. Its reserves term moved -0.211 to -0.894 on the build, and its SOFR-minus-IORB term from +0.40 to -0.20 as the spread crossed zero.11
So the regime label changed while nothing structural did. Scarcity is still +0.975, near the top of its scale, and the reserve stock is $2,930.1bn against a full-history mean near $1,791bn.1,11 The composite is reading a collateral market that healed, not a system that eased.
A repaired funding market and the BTC complex
Bitcoin is $74,558.58, -1,264.91 over 5 sessions and -10.76% over thirty, at the 43.9th percentile of its range. Ether $2,114.87, -114.16 on the week and at the 19.3th.12
The Strategy common is $125.59, off $3.34 and at the 18.8th percentile. Against a net asset value of $164.15 per share - 846,000 coins over 384.23m shares - that is 0.765 times the coins it holds.13
The preferreds split again. STRC $99.98, up $0.18 and at the 100.0th percentile of its range; STRF $103.01, off $0.19, at the 76.0th; STRK $67.40, off $0.45, at the 28.2th.13
A cash coupon at the top of its range against the equity claim at the 19th is the same trade this column has run for three weeks. A funding market back at the administered rate does not fix it: the discount on a levered claim is set by the real 10yr at 2.66%, not by the overnight spread.5,13
What would change the read
The first test is Wednesday's reserve print. The general account has already paid out a further $25.9bn since the reserve date, $936.2bn to $910.3bn on 6 Nov, which points to a third consecutive build. A print below $2,930.1bn against that payout would make the residual structural rather than a timing difference.1,2
The second is the spread. Secured money back above the administered rate for five consecutive sessions would say the premium is a level, not an episode. One negative print does not retire a 10-session run; it is the first confirmation the other way.4
The third is the facility. Reverse repo above $50bn remains the threshold that would put a real buffer under the next build, and $0.252bn is as far from it as a print gets. Anything under $5bn keeps every chequebook move hitting reserves one-for-one.3
The fourth is credit. High-yield OAS back through 3.48% would say the 6.0 basis points were a bounce inside a widening trend; back through 3.10% would say the widening is over and last week's 2026 high was the peak.9
The fifth is the corridor. 2s10s has printed 38.0 on 30 Oct, 35.0 on 6 Nov and 32.0 on 9 Nov. Until it clears 38.0 on the top or 30.4 on the bottom, the curve is not saying anything the funding tape has not already said.5,6
Everything between here and those lines is a reserve base $36.3bn higher, a chequebook $79.4bn lighter with $43.1bn unaccounted for, a parking lot at $0.252bn, and secured money 1 basis point below the rate the Fed pays. That is not a forecast. It is Wednesday's comparison.
Notes
1. Reserve balances $2,930,090m as of 2026-11-04, from $2,893,770m (2026-10-28), $2,962,610m (2026-10-21), $3,029,660m (2026-10-07). Weekly change +36.3bn, first positive print since 7 October; $32.5bn below 21 October, $99.6bn below 7 October. Source: FRED WRESBAL, as of 2026-11-04.
2. Treasury general account closing balances: $910.3bn (6 Nov), $922.8bn (5 Nov), $936.2bn (4 Nov), $938.2bn (3 Nov), $930.8bn (2 Nov), $951.6bn (30 Oct), $1,040.2bn (29 Oct), $1,015.6bn (28 Oct). Payout over the 7 days to the reserve date $79.4bn; a further $25.9bn out of the account to 6 Nov. Source: US Treasury Fiscal Data, Daily Treasury Statement, operating cash balance, as of 6 Nov.
3. Overnight reverse repo take-up: $0.252bn (9 Nov), $0.150bn (6 Nov), $0.125bn (5 Nov), $0.225bn (4 Nov), $0.427bn (3 Nov), $5.087bn (2 Nov), $4.375bn (30 Oct), $0.353bn (29 Oct), $11.539bn (30 Sep). Source: FRED RRPONTSYD, as of 9 Nov.
4. SOFR 3.89% on 2026-11-06 against IORB 3.90% - minus 1 basis point. Positive run of 10 sessions from 2026-10-23 to 2026-11-05: +2, +3, +5, +4, +4, +5, +5, +7, +4, +2. Prints immediately before the run: -2 (2026-10-21), +0 (2026-10-22). EFFR 3.88% on every session from 26 October to 2026-11-06; secured money traded 9 basis points above unsecured on 3 November and 1 on 6 November. Sources: Federal Reserve Bank of New York reference rates, as of 6 Nov; FRED IORB.
5. Treasury par yield curve, 9 Nov close: 3-month 4.24%, 2yr 4.56%, 5yr 4.67%, 7yr 4.76%, 10yr 4.88%, 20yr 5.29%, 30yr 5.26%. Session change (vs 6 Nov): 3-month +5.0bp, 2yr +5.0, 5yr +3.0, 7yr +2.0, 10yr +2.0, 20yr 0.0, 30yr 0.0. Week change (vs 30 Oct): 3-month +8.0bp, 2yr +0.0, 5yr -4.0, 7yr -4.0, 10yr -6.0, 20yr -5.0, 30yr -5.0. Source: US Treasury daily par yield curve, as of 9 Nov.
6. 2s10s on the par close: 32.0bp on 9 November (4.56% less than 4.88%), 35.0bp on 6 November, 38.0bp on 30 October. The site's cash-mark series printed 37.7 at the 6 November snapshot and 30.4 at today's; the par close is quoted here because the Levels row reconciles against it.
7. 10yr breakeven 2.22% (2026-11-09), from 2.23% (6 Nov), 2.24% (5 Nov), 2.26% (4 Nov) and 2.34% (30 Oct); lowest since 2026-08-05. Source: FRED T10YIE, as of 2026-11-09.
8. S&P 7,549.94 (+173.08 session, +8.54 week, 92nd percentile); NDX 27,907.63 (+917.64); VIX 19.92 from 25.64 (19.9th); MOVE 85.98 from 92.77 (-14.38 week, 35.9th); HYG $77.51 (+0.28), JNK $93.01 (+0.31), LQD $104.67 (+0.14), TLT $81.96 (+0.04); gold $4,129.30 (+14.80 session, +191.10 week), silver $53.33, copper $6.325, WTI $85.28 (-13.61 week), Brent $91.70, DXY 101.646, USDJPY 158.00. Cash marks as of 9 Nov; metals, energy and FX as of today's 07:00 UTC snapshot.
9. High-yield OAS 3.42% (2026-11-06), from 3.48% (2026-11-05, the 2026 high) and 3.44% (30 Oct); 6.0bp of tightening in one session. Investment-grade OAS 0.93% (2026-11-06), 2026 high 0.94% (2026-11-02). Sources: FRED BAMLH0A0HYM2, BAMLC0A0CM, as of 2026-11-06.
10. Initial claims 189000 (week to 2026-10-31); continuing claims 1,718,000 (week to 2026-10-24); payrolls 159,293k in 2026-10, +165k on the month; unemployment 4.1% in 2026-10. Core CPI 2.57% y/y, headline 3.72% (September). Sources: FRED ICSA, CCSA, PAYEMS, UNRATE, CPILFESL, CPIAUCSL, as of those dates.
11. Composite headline +0.297 NEUTRAL on 5/5 sub-indices, from +0.496 TIGHTENING. Sub-indices: Rates & Policy +0.662, Liquidity -0.563, Credit & Stress +0.457, Growth & Inflation -0.045, Scarcity +0.975 (dominant). Liquidity terms: RRP -0.595, Reserves -0.894, SOFR-IORB -0.200. Positive is stress. Source: site composite.
12. Bitcoin $74,558.58 (-1264.91 over 5 sessions, -10.76% over 30, 43.9th percentile); ether $2,114.87 (-114.16 week, 19.3th). As of today's 07:00 UTC snapshot.
13. Strategy common $125.59 (-3.34, 18.8th percentile); NAV $164.15 per share on 846,000 coins over 384.23m shares; 0.765x. STRC $99.98 (100.0th), STRF $103.01 (76.0th), STRK $67.40 (28.2th). Source: data/mstr-marks.json, as of today's 07:00 UTC snapshot.
| TENOR | LEVEL | 1D Δ | Z (3Y) | RANGE %ILE |
|---|---|---|---|---|
| 2Y | 4.560 | +0.050 | +1.11 | 71 |
| 5Y | 4.670 | +0.030 | +1.68 | 75 |
| 10Y | 4.880 | +0.020 | +1.86 | 74 |
| 30Y | 5.260 | 0.000 | +1.61 | 76 |