The Gap Between Two-Year and Ten-Year Yields Shrank Because Short Rates Rose, Not Because Inflation Did
The "corridor" is the distance between what the US government pays to borrow money for two years and what it pays for ten. Traders call it 2s10s. It matters because it is the bond market's one-line answer to a simple question: are the next few years about the direction of policy, or about how much extra investors demand for lending long?
A week ago the answer was 35 hundredths of a percentage point, and the long end had done all the work. On Friday's close the answer was 27.4 hundredths, another 7.5 narrower, and the engine had changed ends again. This time it was the two-year that moved hardest: its yield rose 0.37 points to 4.88%, while the ten-year rose 0.295 points to 5.154%. The short end did the compressing.
What matters more is what the rise was made of. A "breakeven" is the bond market's estimate of inflation over the life of a bond. The "real yield" is what is left once you subtract it — the pure cost of money, with inflation stripped out. Last week the entire fall in the ten-year yield came out of the inflation estimate. This week, of the ten-year's 0.295 point rise, 0.285 was real. Inflation expectations contributed one hundredth of a point.
Credit, meanwhile, undid its entire selloff. Three markets, one week, and only one of them is now pricing any kind of squeeze.
What moved
The 13 November close, across the curve: two-year 4.88%, five-year 4.98%, seven-year 5.05%, ten-year 5.154%, twenty-year 5.52%, thirty-year 5.481%, three-month bill 4.32%.
Every point rose, and the order is the reverse of last week's. The two-year gained 0.37 points, the five-year 0.34, the seven-year 0.31, the ten-year 0.295, the twenty-year 0.23, the thirty-year 0.221. The three-month bill, which a week ago was the only maturity to rise, gained the least this time: 0.13 points to 4.32%.
That ordering is the tell. When the three-month bill leads, the move is about cash and the next few meetings. When it lags while everything else rises, the move is about where policy ends up, not about what the Fed does this month.
Where each yield sits in its own three-year range has rotated too. The five-year is now in the 93rd percentile of that range, the seven-year the 92nd, the ten-year the 91st, the two-year the 90th. The thirty-year is at the 88th and the three-month bill only the 37th. A week ago the thirty-year was the top of the file. The five-year's 20-place jump in the ranking was the biggest single move on the curve.
The dollar was 103.48, up 1.74 on the week. Dollar-yen 159.55, up 1.47. Euro 1.10, down 0.02. A firmer dollar alongside a selloff led by the middle of the curve says the move is about the price of dollars, not about growth.
The short end did the flattening
The gap is 27.4 hundredths on the 13 November cash marks, from 34.9 on 6 November. That is a third direction in four weeks: 36.9 on 30 October, 34.9 on 6 November, 27.4 now.
The arithmetic is simple. The two-year went from 4.51% to 4.88% — up 0.37 points. The ten-year went from 4.859% to 5.154% — up 0.295. When the short end rises faster than the long end, the gap between them shrinks. That is the opposite of last Monday, when the long end supplied the narrowing.
An independent source agrees. The published par curve — a separate file built from on-the-run bonds — shows a 24.0 hundredths gap for 12 November against 25.0 the day before and 32.0 on 9 November. That file runs a session behind the cash marks, and the cash gap on 12 November was 24.3. The two measures differ by 0.3 hundredths. There is no disagreement to explain.
The path is a slow grind, not a break: 36.9 on 30 October, 30.4 on 2 November, 34.2 on 3 November, 31.4 on 4 November, 37.7 on 5 November, 34.9 on 6 November, 30.4 on 9 November, 25.3 on 10 November, 24.3 on 12 November, 27.4 on 13 November.
The 24.3 print was the narrowest in the three weeks this file has tracked the cash gap. The par curve's own low for 2026 was 20.0, set on 21 September. The full-year range runs from that 20.0 to 74.0 on 9 February, so at 24 the gap sits in the bottom tenth of the year — and it got there with both ends rising. A curve that flattens while every yield sells off is not investors buying long bonds. It is the near end being repriced harder than the far end.
Almost none of the rise was inflation
The ten-year yield rose 0.295 points. The inflation estimate inside it rose 0.01. The real yield rose 0.285.
The breakeven is 2.24% as of 13 November, from 2.23% on 6 November, and 0.06 points off its 2026 low of 2.18% set on 24 June. The derived real yield is 2.914%, from 2.629%. The Federal Reserve's own published real yield is 2.86% as of 12 November, from 2.63% on 6 November — a 0.23 point rise, and the highest since 13 October.
This is the exact mirror of last week. Then the ten-year yield fell 0.07 points and the inflation estimate fell 0.11, so the whole move came out of inflation and the real yield actually rose. Now the ten-year rises 0.295 and inflation contributes 0.01, so the whole move is real. The bond market has not repriced inflation in either direction for two weeks. It has repriced the price of money twice, and the inflation estimate has sat at 2.2% through both.
Gold agrees this week, which it did not last week. Gold is $3,842.30, down $267 on the week, or 6.5%. Silver fell 10.0% to $47.285. A metal that pays no interest should fall when the real yield rises, and this week it did — after rising 5% into a collapse in inflation expectations a week earlier. The relationship reasserted itself rather than breaking.
Oil drove none of it. West Texas Intermediate is $89.22, down $1.77 on the week; Brent $95.66. An inflation estimate that moves one hundredth of a point while crude falls and gold drops 6.5% is a market that has stopped pricing inflation in either direction. That leaves the real yield as the only live variable in the corridor.
Copper fell 4.9% to $6.027 and the copper-to-gold ratio barely moved, 0.00156 against 0.00154. The industrial metal lost 4.9% while the monetary one lost 6.5%. Both fell because the discount rate rose, not because either is telling a different story about demand.
Credit undid the whole selloff
High-yield credit spreads — the extra yield a junk-rated company pays over a Treasury of the same maturity, which is the market's price of default risk — are 3.20% as of 12 November. They were 3.48% on 5 November. That is 0.28 points tighter in five sessions: the entire widening that had taken spreads to their 2026 high, and the tightest reading since 28 October.
Investment-grade spreads are 0.89%, from 0.94% at the end of October. They sit 0.05 points under their own 2026 high, also 0.94%, set on 16 March.
The bond funds fell anyway. HYG lost $0.29 to $76.94, JNK $0.38 to $92.32, LQD $1.72 to $102.81, and TLT $2.54 to $79.38. A credit fund falling while its spread tightens is not a credit loss — it is a duration loss, the same 0.295 point rise in the ten-year yield passing straight through a long-maturity fund.
Volatility markets split, and in the opposite direction to last week. The VIX, which measures expected swings in US equities, is 19.30 — down 6.34 points and in the 18th percentile of its range. The MOVE index, the same measure for Treasury bonds, is 98.79, up 0.87 and near its median. Rate volatility rose while equity volatility collapsed. On 9 November it was the reverse.
Equities took the other side of the real-yield rise entirely. The S&P 500 is 7,561.67, up 2.5% on the week. The Nasdaq 100 is up 3.5%. A 0.285 point rise in the real ten-year yield alongside a 2.5% gain in the index is not a market discounting that move.
Last Monday's question was which market had it wrong: the curve, pricing too little inflation, or credit, pricing too much default. The answer arrived within five sessions — credit was wrong, and it has retraced. What replaced it is a real-yield repricing that neither credit nor equities have acknowledged.
Money markets stayed quiet, and that matters
Bank reserves were $2.96 trillion in the week to 11 November, up $30.5 billion on the week and following a $36.3 billion rise the week before.
The Treasury's cash balance at the Fed was $933.1 billion on 12 November, up $10.3 billion on the week. A rising Treasury balance normally drains reserves, so the two cannot both rise without an offset elsewhere on the Fed's balance sheet. This week they did both rise, which means other items moved; the two lines are not the whole story.
The overnight repo facility, where money market funds park spare cash at the Fed, took $0.325 billion. It has been a rounding error for a month.
The most useful line is the funding spread. Overnight secured borrowing fixed at 3.90%, exactly the rate the Fed pays on reserves — a zero premium, down from 0.02 points on 5 November. Effective fed funds was 3.88%.
That combination is worth holding onto: the short end sold off 0.37 points in a week when nothing in the funding plumbing was under stress. The move came from expectations about the policy path, not from a scramble for collateral.
The labour data is quiet too. Initial jobless claims were 200,000 in the week to 7 November, from 191,000. Continuing claims were 1.697 million in the week to 31 October, from 1.717 million. A 9,000 rise in new claims alongside 20,000 fewer people still claiming is not a labour market cracking.
The model's aggregate reading is 0.418, on a scale where positive means financial conditions are tightening. The dominant component is scarcity at 1.45, up from 0.945 in the previous snapshot. Credit and stress is 0.352, rates and policy 1.168, growth and inflation −0.376, liquidity −0.504. Because the consumer price data is missing from this snapshot, growth and inflation is being scored only on claims, the inflation estimate and the copper-gold ratio. Scarcity is the only component above 1, and it is driven by the real ten-year yield — the same variable that produced 0.285 of the ten-year's 0.295.
Bitcoin fell 13%, and the company holding it fell 13.4%
Bitcoin is $64,668, down $9,541 on the week, or 12.9%, and down 20.8% over thirty sessions. Ether is $1,815.70, down 11.8%.
Strategy (MSTR) holds 846,000 coins against 384.23 million shares. At a $64,672 bitcoin price that is $142.39 of coins per share, so the equity at $111.62 trades at 78 cents for every dollar of bitcoin it owns — a 21.6% discount to its holdings. A week ago the same arithmetic gave 0.789, so the discount widened by about half a point.
The coin fell 12.9% and the wrapper fell 13.4%. The multiple barely moved, and in a week when the real ten-year yield rose 0.285 points, a 21.6% discount to held coins is not behaving like a carry trade — because it is not one.
The preferred shares split three ways. STRC is $99.91, up $0.07 and at the top of its range. STRF is $100.35, down $2.85. STRK is $66.16, down $1.69. A 2.92 point spread between the best and worst of the three in a single week says the market is pricing each instrument on its own terms, not the structure as one credit.
What would change this read
First, inflation expectations waking up. If the ten-year breakeven rose above 2.34% — the 30 October level — with the ten-year yield at or above 5.154%, the rise would be about inflation compensation again. That is the debasement story, and it is not what a one-hundredth-of-a-point move looks like.
Second, the middle of the curve giving it back. If the five-year fell back below 4.64%, its 6 November level, while the two-year held at or above 4.88%, then the 13 November move was a week of heavy five-year supply rather than a repricing of where policy ends up — and the gap would widen again from the long end with no change in the policy path.
Third, credit breaking the other way. If high-yield spreads widened through 3.48% with the ten-year holding 5.154%, the real-yield rise would look like a default event and last week's read would be back on the table. If they tightened through 3.14%, the 28 October level, with the ten-year still at 5.154%, the whole episode is supply and term premium and nothing else.
Between those three lines the conclusion is narrow. The gap narrowed because the two-year rose 0.37 points against the ten-year's 0.295, and 0.285 of that 0.295 was real yield. Credit retraced 0.28 points to 3.20%. Equities added 2.5%.
The bond curve is pricing a higher real cost of money. Credit and equities are pricing nothing at all. Last week the curve was the odd one out, and it was wrong. This week it is the odd one out again — and the invalidation is a ten-year breakeven above 2.34%.