The State Cleared the Overnight Premium. The 2yr Rose 40bp Anyway.
One market sets the price of cash for one night. Another sets the price of cash for two years. They are not the same market, they do not clear against the same balance sheet, and this week they moved in opposite directions. The overnight repo rate - SOFR, the rate at which cash is swapped for Treasury collateral and returned tomorrow - fell below the rate the Fed pays on reserves for the first time in eleven sessions. The two-year Treasury yield, which is the market's price for holding the state's paper across two years of policy, rose 40 basis points in four steps. Both are prints of the same sovereign. Only one of them is under the sovereign's direct control. This column has spent three Fridays on the gap between the administered rate and the traded one, because that gap is the only public read on whether the system has spare cash. This week the gap closed, which should have been the end of the story. It was the beginning of a different one: what the chequebook can and cannot buy.
The chequebook cleared the premium and the residual changed sign
The Treasury's general account closed at $912.1bn on 10 November, from $936.2bn on 4 November - $24.1bn out in four sessions, and $140.4bn below the $1,052.5bn record close of 27 October.1
Reserve balances rose $30.5bn over the corresponding week to $2,960.6bn (week ended 11 November), from $2,930.1bn (week ended 4 November).2 That is the highest reserve print since $2,962.6bn on 21 October.
State the arithmetic before drawing anything from it. A week ago the account released $79.4bn and $36.3bn of it reached reserves - 46% pass-through, with $43.1bn going somewhere the two series tracked here do not see. This week the account released $24.1bn and reserves gained $30.5bn, so $6.4bn arrived in the bank base that the chequebook did not supply. Currency in circulation and the foreign repo pool are the usual candidates in both directions, and this column measures neither.
The honest reading is narrow, and it is narrower than last week's. Pass-through went from 46% to 127% - the link between the chequebook and the bank base re-tightened rather than broke. But it missed in both directions inside two weeks, by $43.1bn one way and $6.4bn the other, so the chequebook is a bound on the sign of the move and not an estimate of its size. That is the same conclusion Tuesday's read reached, with a smaller error.
Three negative fixings after ten positive ones
Secured overnight money fixed 3.89% on 10 November against an administered 3.90%. That is minus 1bp, the third consecutive negative fixing, after ten consecutive positive ones to 5 November.3
The run is the whole of the reversal in one frame. From 26 October the premium ran +3, +5, +4, +4, +5, +5, +7, +4, +2 basis points, peaking at +7 on 3 November, then -1, -2, -1 on 6, 9 and 10 November.3 The 120-fixing window from 20 May 2026 holds longer positive runs than the ten just ended - 17 fixings from 2 to 28 July, 13 from 8 to 25 September, 10 from 7 to 22 October - so the length of the run is not the signal. The speed of the reversal is: ten sessions of premium erased in three.
Volume fell with the premium rather than against it. $3,159bn traded on 10 November against $3,268bn on 2 November, the window high.3 The effective fed funds rate held at 3.88% throughout, so the unsecured market never shared the secured market's premium and does not share its collapse. This was collateral, not policy.
Read together with the chequebook, the sequence is legible. The state spent $140.4bn out of its account in ten sessions and the collateral premium went from +7bp to -1bp. That is the cleanest demonstration of the mechanism this column has recorded: the quantity of overnight cash is a policy variable and the price of overnight cash follows it.
The 2yr took 40bp and no other tenor came close
The 2yr par yield closed at 4.86% on 12 November, from 4.46% on 5 November. That is +40bp in four steps, and it is the largest four-step rise in the 217 par sessions this file holds for 2026; the next largest is 26bp, in September.4
The distribution matters more than the level. Every tenor from 2yr to 30yr rose, and the rise declines monotonically with maturity: +36bp at 5yr, +32bp at 7yr, +26bp at 10yr, +21bp at 20yr, +18bp at 30yr. The 3-month bill rose +15bp.4 This is not a long-end supply event and it is not a short-end cash event. It is a repricing of the policy path across the two-to-five year sector, where the market's expectation of the next several meetings is priced.
The curve did what that shape implies. 2s10s narrowed from 38.0bp on 5 November to 24.0bp on 12 November - 14bp of bear flattening - and there is exactly one date in 2026 with a lower slope: 21 September, at 20.0bp.4 A flattener of that speed with the front end rising is the market withdrawing rate-cut credit, not adding term premium.
This is the week's governance content. On 5 November the state's overnight cash was scarce enough to command a 4bp premium. By 12 November the state had spent its chequebook, cleared that premium to minus 1bp, and the 2yr had risen 40bp anyway. The instrument that sets the overnight price cannot set the two-year price. One is a settlement balance. The other is a claim on twelve future policy decisions.
Every basis point of the 10yr move was real
The 10yr nominal went from 4.84% on 5 November to 5.10% on 12 November, +26bp. The breakeven - the gap between the nominal yield and the yield on inflation-protected paper, which is the market's price of expected inflation over the life of the bond - printed 2.24% on 12 November against 2.24% on 5 November.5
Subtract one from the other and the real yield went from 2.597% to 2.863%, +26.6bp.5 The entire move was real. Not one basis point of the repricing was the market charging more for inflation; the whole of it was the market charging more for the money itself.
That distinction decides what kind of event this is. A nominal selloff built from breakevens is a monetary-credibility story and it hits the currency first. A selloff built from real yields is a cost-of-capital story and it hits every asset that discounts a future cash flow. Gold is the cleanest test available in this file and it failed the credibility test: $3,894.90, down $172.10 on the week.6 Gold does not fall 4.2% in a week when the market is losing faith in a central bank's inflation control. It falls when the real yield it competes with goes up - which is what happened.
Credit tightened 12bp and the all-in cost rose 14
High-yield spreads narrowed. Option-adjusted spread on the high-yield index printed 3.30% on 11 November, from 3.42% on 6 November - 12bp tighter. Investment grade went from 0.93% to 0.90%, 3bp tighter.7
The all-in cost of borrowing rose anyway, because the risk-free leg did more work than the spread leg. High yield at 3.30% over a 10yr at 5.10% is an 8.40% all-in, against 8.26% on 6 November when the same two legs were 3.42% and 4.84%: +14bp.7 Investment grade is 6.00% against 5.77%, +23bp, because its spread is a smaller share of the total and the duration leg dominates it.
The funds agreed with the arithmetic and not with the spreads. HYG closed at 76.98, down 0.37 on the session; JNK at 92.42, down 0.41; LQD at 103.05, down 0.91 and down 1.61 on the week; TLT at 79.72, down 1.20 and 2.21 on the week, in the 19th percentile of its own history.7 A tightening spread and a falling bond fund are not a contradiction. They are two prices of the same credit at two different durations, and the duration one is losing.
This is the sequence that would matter if it persisted. Credit is not yet pricing stress - 3.30% is 18bp inside the 3.48% print of 5 November - but the borrower's cost is rising while the borrower's risk premium is falling. Only one of those can keep going.
The 20yr moved 5bp over the 30yr
The 20yr closed at 5.48% on 12 November and the 30yr at 5.43%. The 20yr now sits 5bp above the 30yr, from 2bp on 5 November.4
That is a small number with a specific meaning. When the 20yr trades through the 30yr, the market is asking to be paid more for the twentieth year of a bond than for the thirtieth - which is not a growth view or an inflation view but a preference about receiving long-dated paper at all. The 30yr has printed above the 20yr for most of 2026; the gap was last at 5bp on 19-21 October, and it has widened in four of the last five sessions.4
Both sit well above the 10yr at 5.10%, so the curve is upward-sloping from 2yr to 20yr and then mildly inverted at the back. That is the shape of a market that will take duration at a price and prefers the shorter end of the long end. It is also the shape that makes the sovereign's own issuance decision expensive in exactly the maturity where it has been borrowing least: the last 90 days settled $5,742.5bn of bills against $971.5bn of coupons, an 85.5% bills share.8
Bill-heavy issuance is cheap at the front and does nothing for the 20yr. The 20yr is priced by buyers who are not being sold to.
Equity risk took the same week at +3.25%
The S&P closed at 7,592.40, down 26.98 on the session and up 238.67 on the week, +3.25%, in the 93rd percentile of its own history. The Nasdaq 100 closed at 28,289.19, up 1,659.38 on the week, +6.23%.9
The volatility market is the part worth reading. The VIX at 18.70 is up 1.21 on the session and down 9.48 on the week, in the 17th percentile of its history. The MOVE - the rate market's equivalent, an index of implied Treasury-option volatility - is 98.43, up 6.00 on the session and up 4.55 on the week, in the 51st percentile.9
So the two-year yield rose 40bp and rate volatility rose with it, while equity volatility fell to the 17th percentile and equities added 3.25%. Duration repriced and risk appetite did not. That is the difference between a discount-rate event and a growth event, and the tape is pricing the first.
Do not overstate it. The S&P's 93rd percentile is a level, not a signal, and a market that has risen into a 40bp front-end selloff is a market that has decided the selloff is about the policy path rather than about earnings. That decision can be reversed by any single payroll print. It has not been reversed yet.
The contract priced at par, the residual at the 16th percentile
STRC - the preferred instrument in the MicroStrategy structure, a perpetual with a stated dividend and a claim ahead of the common - closed at 99.96 on 12 November, in the 100th percentile of its 326-session history. MSTR closed at 114.41, down 7.94 on the session, down 14.89 on the week, -11.5%, in the 16th percentile of 753 sessions.10
The gap between those two percentiles is the read. A 100th-percentile preferred against a 16th-percentile common is a market that has not changed its view of the contractual claim and has changed its view of the residual one by the full amount of the drawdown. STRF at 101.81 sits in the 73rd percentile; STRK at 67.15 in the 28th.10 The further up the capital stack, the better the price held.
The underlying arithmetic explains it without needing a view. The company holds 846,000 BTC against 384.23m shares, which is 0.0022 BTC per share; at a $69,005 coin that is $151.98 of net asset value per share, against a $114.41 close. mNAV is 0.7528 - the common trades at a 24.7% discount to the coin it holds.10
That discount is the governance fact. The structure's equity is no longer being valued as a claim on Bitcoin at a premium; it is being valued at a discount to the coin, while the instrument with the contractual coupon is at par. When a wrapper's residual claim trades below the thing it wraps, the market is charging the wrapper for its complexity and its financing, not for its assets. Bitcoin itself is $69,005, -8.7% on the week and in the 36th percentile; ETH is $1,937.18, -9.8% and in the 13th.11
Where this lands on the tape
The composite is +0.361, regime TIGHTENING, dominant sub-index Scarcity at 1.30, coverage 5 of 5. A session ago it was +0.329 and NEUTRAL.12
The move is in the two components that should have moved. Rates & Policy went from +0.865 to +1.070 on the 2yr and 10yr levels and a flatter slope; Liquidity went from -0.414 to -0.571 on reserves and the SOFR-IORB leg. Credit & Stress rose from +0.308 to +0.407 on the MOVE.12
One line for the record: the composite turned TIGHTENING on a week when the state spent $24.1bn, the bank base gained $30.5bn, and the overnight premium went negative. Liquidity loosened in every series this column measures and the index tightened anyway, because the index is weighted to price, and price moved.
Where this breaks
The separation read dies if the 2yr retraces the 40bp while SOFR stays at or below 3.90%. That combination would mean the front end move was a positioning event that the funding market never corroborated, and the two markets would be one market again.
The real-yield read dies if the breakeven moves more than 5bp from 2.24% on the next print. The claim that the entire 10yr move was real requires the inflation leg to stay where it is; a breakeven above 2.29% puts a third of the move back into inflation compensation.
The credit read dies if high yield prints above 3.48% - the 5 November level - while the 10yr holds above 5.10%. Spreads widening into a rising risk-free rate is the combination that turns a discount-rate event into a default event, and it is the one combination this week's tape does not show.
The structure read dies if STRC falls below 99.90 with the common above $125. That would make the 100th percentile a flow - a single large buyer of the preferred - rather than a ranking, and the capital-stack ordering would stop describing anything.
The confirming case is all four holding at once: SOFR at or below 3.90% into the next fixing window, the 2yr above 4.86% or within 5bp of it, the breakeven at 2.24%, and high yield under 3.48%. Four prices, one actor, and the actor is not the one setting the overnight rate.
Notes
1 Treasury general account, closing balance, Treasury Fiscal Data operating cash balance: $912.1bn (10 Nov), $922.6bn (9 Nov), $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), $1,052.5bn (27 Oct). The 27 October close is the highest in the 100-session window this column holds, which begins 22 June 2026; the window low is $738.3bn on 10 July.
2 Reserve balances $2,960.6bn (FRED WRESBAL, week ended 11 Nov), from $2,930.1bn (week ended 4 Nov), $2,893.8bn (week ended 28 Oct), $2,962.6bn (week ended 21 Oct), $3,022.1bn (week ended 14 Oct). Change +$30.5bn, or 127% of the $24.1bn the chequebook released between 4 and 10 November; the same calculation a week earlier gave 46% of a $79.4bn release, leaving $43.1bn unaccounted for. This week's residual runs the other way: $6.4bn reached the bank base that the chequebook did not supply.
3 SOFR and EFFR from NY Fed reference rates; IORB from FRED (3.90% on every date shown). Fixings: 10-26 3.93 (+3), 10-27 3.95 (+5), 10-28 3.94 (+4), 10-29 3.94 (+4), 10-30 3.95 (+5), 11-02 3.95 (+5), 11-03 3.97 (+7), 11-04 3.94 (+4), 11-05 3.92 (+2), 11-06 3.89 (-1), 11-09 3.88 (-2), 11-10 3.89 (-1). Volumes $bn: 10-26 3,086, 10-27 3,147, 10-28 3,074, 10-29 3,145, 10-30 3,175, 11-02 3,268, 11-03 3,260, 11-04 3,207, 11-05 3,256, 11-06 3,174, 11-09 3,100, 11-10 3,159. The 120-fixing window from 20 May 2026 contains longer positive runs: 17 (2-28 Jul), 13 (8-25 Sep), 10 (7-22 Oct).
4 Par yield curve, Treasury daily rates: 12 Nov 3M 4.32, 2yr 4.86, 3yr 4.90, 5yr 4.95, 7yr 5.02, 10yr 5.10, 20yr 5.48, 30yr 5.43; 2s10s 24.0bp; 20yr over 30yr 5bp. 5 Nov: 3M 4.17, 2yr 4.46, 5yr 4.59, 7yr 4.70, 10yr 4.84, 20yr 5.27, 30yr 5.25; 2s10s 38.0bp; 20yr over 30yr 2bp. The 217-session file begins 2 January 2026. Four-step 2yr rises, largest first: +40.0 (5-12 Nov), +26.0 (8-14 Sep), +26.0 (4-11 Sep), +24.0 (9-15 Sep). Dates in 2026 with 2s10s at or below 24.0bp: 21 Sep (20.0) and 12 Nov (24.0). 20yr over 30yr last at 5bp on 19, 20 and 21 October.
5 10yr real yield 2.863% (12 Nov), from 2.597% (5 Nov), both computed as the nominal par yield less the FRED T10YIE breakeven; the breakeven printed 2.24% on both dates (FRED T10YIE, 12 Nov). FRED DFII10, the direct real-yield series, lags at 2.78% (10 Nov) because its source file has not refreshed past that date; the computed figure is used and labelled as computed.
6 Gold $3,894.90 (+$9.60 session, -$172.10 week, 58.0); silver $49.745 (-$0.227, -$2.058 week); copper $6.1730 (-$0.0095); WTI $88.93 (+$0.19, -$6.08 week); Brent $95.73 (-$0.09, -$2.10 week); DXY 103.170 (+0.870 week); USDJPY 159.762 (+2.453 week); EURUSD 1.1061 (-0.0168 week).
7 HY OAS 3.30% (FRED BAMLH0A0HYM2, 11 Nov), 3.42% (6 Nov), 3.48% (5 Nov); 1yr window 2.60% to 3.48%. IG OAS 0.90% (BAMLC0A0CM, 11 Nov), 0.93% (6 Nov); 1yr window 0.73% to 0.94%. All-in high yield = OAS + 10yr par: 8.40% (11-12 Nov) against 8.26% (6 Nov). All-in investment grade 6.00% against 5.77%. Marks 12 Nov: HYG 76.98 (-0.37 day, -0.30 week, 89.6), JNK 92.42 (-0.41, -0.36, 89.5), LQD 103.05 (-0.91, -1.61, 74.2), TLT 79.72 (-1.20, -2.21, 19.2).
8 Treasury auction settlements, 90 days to 12 November, Treasury Fiscal Data: gross $6,713.9bn, of which bills $5,742.5bn (62 auctions), notes $842.5bn (16), bonds $128.9bn (7); bills share 85.5%. Prior 90 days to 13 August: gross $8,182.1bn, bills $7,104.5bn (86.8%), coupons $1,077.6bn.
9 Marks 12 Nov: SPX 7,592.40 (-26.98 day, +238.67 week, +3.25%, 93.4); NDX 28,289.19 (-85.06, +1,659.38, +6.23%, 81.4); QQQ 688.17 (-2.21, +41.12, 81.6); VIX 18.70 (+1.21, -9.48, 16.9); MOVE 98.43 (+6.00, +4.55, 50.7). Percentile ranks are against each series' own stored history.
10 STRC 99.96 (+0.01 day, +0.6393 week, 100.0 of 326), MSTR 114.41 (-7.94, -14.89, -11.5%, 16.2 of 753), STRK 67.15 (-1.14, 27.7 of 448), STRF 101.81 (-0.54, 72.9 of 412). Structure marks: 846,000 BTC held, 384.23m shares, 0.00220181 BTC per share, NAV $151.98 per share, mNAV 0.7528 at a $69,026 coin mark. The 24.7% discount is 1 - 0.7528.
11 BTC $69,005.14 (-$4,959.33 day, -$6,587.33 week, -8.71%, -16.63% 30d, -8.83% 5d, 36.4); ETH $1,937.18 (-$118.70, -$210.85, -9.82%, 13.0).
12 Composite +0.361, regime TIGHTENING, dominant Scarcity 1.300, coverage 5/5. Sub-indices: Rates & Policy +1.070, Liquidity -0.571, Credit & Stress +0.407, Growth & Inflation -0.401, Scarcity +1.300. A session earlier: headline +0.329, NEUTRAL, Rates & Policy +0.865, Liquidity -0.414, Credit & Stress +0.308.
13 Initial claims 200k (FRED ICSA, week ended 7 Nov) from 191k (week ended 31 Oct); continuing claims 1,697k (CCSA, week ended 31 Oct) from 1,717k (week ended 24 Oct). Core and headline CPI y/y are null in this snapshot: CPIAUCSL and CPILFESL carry an October 2026 level but no 2025-10-01 observation, so the year-ago base is missing and no y/y can be computed. Both are omitted rather than estimated.
| TENOR | LEVEL | 1D Δ | Z (3Y) | RANGE %ILE |
|---|---|---|---|---|
| 2Y | 4.800 | -0.030 | +1.73 | 87 |
| 5Y | 4.900 | -0.030 | +2.38 | 90 |
| 10Y | 5.100 | -0.020 | +2.63 | 89 |
| 30Y | 5.460 | 0.000 | +2.19 | 88 |