Thu: the 5-year TIPS drew 2.41 cover, nothing triggered, and the 10-year fell 19 basis points as the Nasdaq dropped 2.6 percent.
Written early Friday October 23, covering Thursday October 22; the plan was committed at 11:48 AM ET Thursday, before the 1:00 PM auction. The $26 billion 5-year TIPS cleared at 2.367 percent with 2.41 cover and 62.0 percent indirect (TreasuryDirect), inside the no-call band. Grade: nothing triggered. The 10-year fell 19.3 basis points to 4.953 percent, mostly real yields (Treasury); the S&P fell 1.21 percent, falsifying Monday's S&P statement. Claims 194,000 (DOL); gas storage +77 Bcf to 3,644 (EIA).
Written Friday October 23 before the US open (about 6:30 AM ET), covering Thursday October 22. The plan graded below was committed at 11:48 AM ET Thursday, before the 1:00 PM ET TIPS auction; jobless claims and gas storage had already printed and are reported, not graded. Every Thursday price here comes from Thursday's session; Friday's overnight trading is used only to flag a futures roll. The $26 billion 5-year TIPS new issue drew a bid-to-cover of 2.41, inside the range where the plan made no call. Grade: nothing triggered. The bigger story was the rest of the tape: the 10-year yield fell 19.3 basis points to 4.953 percent, the Nasdaq fell 2.6 percent, Brent rose 2.9 percent and gold fell $120. That combination falsified one of the conditional statements in Monday's note.
This continues Wednesday's note, where the 10-year rose 2.9 basis points on breakevens and natural gas fell to 2.775. Thursday took back more than six times that rise, and this time it was mostly real yields.
What the data said
- 5-year TIPS, 1:00 PM ET. $26 billion of a new 5-year TIPS (CUSIP 91282CRR4, series AE-2031, matures October 15, 2031) at a high yield of 2.367 percent and a 2.250 percent coupon, bid-to-cover 2.41. Indirect bidders took $15.93 billion of $25.68 billion in competitive awards (62.0 percent), direct bidders $8.33 billion (32.4 percent) and primary dealers $1.42 billion (5.5 percent). The median yield was 2.300 percent, the low 2.245 percent, and 68.33 percent of bids at the high yield were filled (source: TreasuryDirect auction record). Against the six prior 5-year TIPS auctions listed in the plan, the cover was the second lowest (only April 2025's 2.28 was lower) and the high yield was the highest by 41 basis points (June 2026 cleared at 1.955 percent). The indirect share was close to last October's 62.1 percent. We have no checkable when-issued yield, so we state no tail. For reference only: Treasury's end-of-day 5-year real yield was 2.33 percent (Treasury real yield curve), but nominal yields kept falling after 1:00 PM, so that gap is not a tail either.
- Initial jobless claims, 8:30 AM ET (week ending October 17). 194,000 seasonally adjusted, down 6,000 from the previous week's level, which was revised up to 200,000. The 4-week average was 197,750. Continuing claims for the week ending October 10 rose 25,000 to 1,728,000; the insured unemployment rate held at 1.1 percent (source: Department of Labor weekly claims release, October 22, 2026). investingLive put expectations at 200,000 initial and 1.720 million continuing; we did not check another source. Low claims and a 19 basis point bond rally on the same day do not fit a labour-market story for the rally.
- EIA natural gas storage, 10:30 AM ET (week ending October 16). Working gas rose 77 Bcf to 3,644 Bcf, 35 Bcf (1.0 percent) above the 2021-25 five-year average of 3,609 Bcf and 152 Bcf below a year earlier. Every region added gas; South Central and Midwest each added 28 Bcf (source: EIA Weekly Natural Gas Storage Report). We found no consensus from a source we could check, so we do not call it a build above or below expectations.
The tape
| Market | Wed Oct 21 (4 PM / close) | Thu Oct 22 (4 PM / close) | Change |
|---|---|---|---|
| 10Y yield | 5.146% | 4.953% | -19.3 bp |
| 30Y yield | 5.439% | 5.287% | -15.2 bp |
| 5Y yield | 4.972% | 4.768% | -20.4 bp |
| DXY (4 PM ET) | 101.693 | 101.628 | -0.07 |
| Gold (GC=F, 4 PM ET) | 4,159.5 | 4,039.8 | -119.7 (-2.9%) |
| USD/JPY (4 PM ET) | 158.13 | 156.82 | -0.83% |
| EUR/USD (4 PM ET) | 1.1237 | 1.1216 | -0.0021 |
| Brent (BZ=F, December, 4 PM ET) | 109.73 | 112.87 | +2.86% |
| Natural gas (NG=F, November, 4 PM ET) | 2.775 | 2.777 | +0.1% |
| S&P 500 | 7,757.17 | 7,663.43 | -1.21% |
Sources: Yahoo Finance query1, using the plan's conventions: hourly 4:00 PM ET bar for DXY, gold, USD/JPY, EUR/USD, Brent and natural gas; daily close for ^TNX, ^TYX, ^FVX and ^GSPC. The Nasdaq fell 2.60 percent to 26,008.86 and the Dow rose 0.02 percent to 52,867.98 (Yahoo ^IXIC, ^DJI). Investrade reported settlements of $113.17 for Brent (up $4.35), $99.47 for WTI (up $2.32, a five-month high) and $4,060.40 for December gold (down $89.60), and called the 10-year's fall its biggest daily drop since August 1, 2025. We did not verify the settlements or that comparison. Note for anyone pulling the same series today: Yahoo's continuous NG=F and BZ=F contracts rolled overnight, to December gas and January Brent, so Thursday's bars above are November gas and December Brent.
Grading the plan
The plan had one leg. It classified the 5-year TIPS bid-to-cover: weak at 2.27 or lower (5-year nominal yield closes above its writing-time level of 4.834 percent), strong at 2.63 or higher (closes below it), no call from 2.28 to 2.62, the range of the six prior 5-year TIPS auctions. DXY carried no call in any row. Classification: bid-to-cover 2.41, neither.
| Line in the map | Expected | Actual | Grade |
|---|---|---|---|
| Weak (≤ 2.27): ^FVX close above 4.834% | if triggered | cover 2.41 | Not triggered |
| Strong (≥ 2.63): ^FVX close below 4.834% | if triggered | cover 2.41 | Not triggered |
| Neither (2.28 to 2.62) | no call | cover 2.41; ^FVX 4.768% | No call |
Grade: nothing triggered. No hits, no misses, nothing void. The plan was built so most outcomes would be no call, and for the second day running that is what happened. One thing worth saying plainly: the cover sat in the lower half of the range on a day the 5-year rallied 6.6 basis points from the plan's writing-time level. Had the weak line triggered, it would have graded a miss. The auction was not the thing moving yields.
Event-window record, as the plan specified (not graded): ^FVX opened the 12:50 PM bar at 4.782 percent and the 1:20 PM bar at 4.782, then closed that bar at 4.773 (Yahoo 30-minute bars, which sit on :20 and :50). Nothing measurable at 1:00 PM. FRED's 5-year breakeven for Thursday is 2.45 (FRED T5YIE), up from 2.40.
Real yields did most of the falling
On Treasury's own curves, the 5-year nominal yield fell 17 basis points to 4.78 percent and the 5-year real yield fell 22 to 2.33 percent (nominal, real). The 10-year nominal fell 17 to 4.96 and the 10-year real 12 to 2.64. The 2-year fell 16 to 4.65. So at five years, the real yield fell more than the nominal one and the breakeven rose; at ten years, it was the other way round. FRED shows the same split: T5YIE up 5 basis points to 2.45, T10YIE down 5 to 2.32, and the 5-year, 5-year forward inflation rate (T5YIFR) down 15 to 2.19 (FRED, via our proxy). Near-term inflation compensation rose with oil; longer-run compensation fell.
A fall that large in short-dated real yields is consistent with a lower expected policy path, though it does not prove one. The one named market figure we have is secondhand: investingLive wrote that futures priced "a 22% chance of no further Fed hike this year," and that "another increase" remained "the more likely outcome, but traders were giving a pause more consideration." We did not read CME FedWatch ourselves, and we have no earlier figure from the same source to compare it with, so we do not quantify the shift.
When it happened
Most of the bond move was done before New York's data. ^TNX's first 30-minute bar of the day, at 8:20 AM ET, opened at 5.040 percent, already 10.6 basis points below Wednesday's close, and the yield then fell steadily to 4.953 by 3:00 PM, with no single half hour standing out (Yahoo ^TNX). Gold's slide also started in Asia: GC=F fell from 4,159.0 at 6:00 PM Wednesday to 4,107.3 by 11:30 PM and kept falling through the New York session. Brent climbed from about 110 to 112.5 overnight, eased in the morning, then rose from 111.31 at 11:30 AM to 113.23 by 2:00 PM. The S&P 500 lost most of its ground in the first 30 minutes after the open (7,695.96 to 7,658.58) and drifted lower until a late bounce.
What the reported causes were, by named source and not verified by us:
- Tech and AI spending. Investrade wrote that the Nasdaq's fall was "led by shares of META after earnings results," and that Meta raised its capital spending range to $135 billion to $145 billion from $130 billion to $145 billion. investingLive put Meta down about 7 percent and the Nasdaq 100 down 2.90 percent.
- Oil. Investrade attributed Brent's gain to "Houthi attacks in Saudi Arabia." investingLive referred to reports of possible renewed strikes on Iran, describing them as "reported risks, rather than confirmation." We have not verified either.
- Gold. Investrade: gold fell "to its lowest level in more than two months" as "higher oil prices fueled inflation concerns." investingLive noted the fall "stood out because Treasury yields were sharply lower."
Our read, as a hypothesis: Thursday looked like an equity-led haven bid in bonds, which is what Monday's note said had stopped happening. Gold falling $120 while real yields fell 12 to 22 basis points is the part that does not fit any simple story, and we do not have one. The dollar index barely moved (-0.07) while the yen gained 0.8 percent, consistent with a haven bid that went to the yen and to Treasuries but not to the broad dollar.
Open statements: one falsified
- The S&P 500 condition: falsified. Monday's note said that if the S&P fell 1.0 percent or more on any day from October 21 to 23, the 10-year would close no more than 1 basis point lower that day, because equity selling was no longer producing a bond bid. It was wrong if the 10-year closed 3 basis points or more lower. Thursday the S&P fell 1.21 percent and the 10-year closed 19.3 basis points lower. The statement was wrong, by a wide margin. Equity selling produced a large bond bid.
- The 10-year real yield at or above 2.84. Treasury's real curve shows 2.64 for Thursday, 20 basis points short; FRED's DFII10 for October 22 had not posted at writing. One day left in the window. Not met so far.
- The 10-year breakeven below 2.38. T10YIE 2.32, down from 2.37, with Brent above $100 throughout. The October 15 statement is falsified only by a print of 2.38 or higher; Friday is the last day.
What we do not know
- Why the 10-year fell 10 basis points before 8:20 AM ET. The Meta results and oil headlines are named by others; we have not tied the overnight timing to either.
- Why gold fell with real yields. Investrade's inflation explanation does not square with lower real yields. investingLive offered "position reductions and demand for cash" as plausible. Neither is evidence.
- The TIPS tail. No checkable when-issued yield.
What would change the read
Friday's plan (editorial/plans/2026-10-23.md) was committed Friday at about 6:15 AM ET, before the US data. It grades one thing: the S&P Global flash US composite PMI for October at 9:45 AM ET on Friday October 23 (date from secondary listings; S&P Global's calendar did not open for us), against September's final 58.4. If the flash composite is 55.4 or lower, then the 5- and 10-year yields close Friday below Thursday's 4.768 and 4.953 percent; if 61.4 or higher, above them; each line is invalidated if that yield closes on the other side. Between 55.5 and 61.3 there is no call. The final October Michigan survey is due at 10:00 AM ET (University of Michigan Surveys of Consumers; preliminary sentiment 46.3) and is recorded, not graded. The Fed stays in its blackout through Thursday October 29, with the FOMC statement on Wednesday October 28 at 2:00 PM ET.
What to read alongside
Nothing on this site is investment advice or a recommendation to trade. Most retail traders lose money. Conditional statements published here are falsifiable hypotheses, graded in public, not signals.