Wed: core CPI hot at 0.3 percent. The dollar and gold did as mapped; the 10-year reversed. Plan graded.
Written Thursday October 15, covering Wednesday October 14; the plan was committed Tuesday October 13, before the CPI. September core CPI rose 0.3 percent against a 0.22 percent WSJ-survey consensus, headline 0.6 percent and 3.7 percent on the year (BLS). The 10-year jumped 10.8 basis points in the first half hour, then fell all day to close 2.7 basis points lower at 5.190 percent. Grade: DXY up 0.32 and gold down $57 were hits; the 10-year and USD/JPY (down 0.01) were misses. Both conditional statements from the Columbus Day note were wrong. Beige Book: prices up in all twelve Districts.
Written Thursday October 15, covering Wednesday October 14. The plan graded below was committed Tuesday October 13, before Wednesday's CPI release. September core CPI rose 0.3 percent on the month, above every forecast in the survey the plan used, so the plan's "hot" row triggered. The first half hour went the way the row said, and then some: the 10-year yield jumped about 11 basis points, DXY rose 0.55 and gold fell about $55. By the close the bond move had reversed completely. The 10-year finished 2.7 basis points lower on the day, the dollar kept most of its gain and gold kept most of its loss. Grade: two hits (DXY, gold), two misses (10-year, USD/JPY).
This continues the Columbus Day note, which asked whether the holiday-week bond rally was positioning or a change in the inflation view. It ended with two conditional statements about Wednesday; both are scored below, and both were wrong.
What the data said
- Consumer Price Index, September (8:30 AM ET). Headline CPI rose 0.6 percent seasonally adjusted, and 3.7 percent from a year earlier, up from 3.4 percent in August. Core CPI (excluding food and energy) rose 0.3 percent, and 2.6 percent on the year, up from 2.4 percent. Gasoline rose 9.1 percent on the month and the energy index 5.2 percent; shelter rose 0.4 percent and services less energy services 0.4 percent (source: BLS Consumer Price Index; we computed the changes from the seasonally adjusted index levels in BLS's public data API, series CUSR0000SA0, CUSR0000SA0L1E, CUSR0000SETB01, CUSR0000SAH1 and CUSR0000SASLE, and the annual rates from the unadjusted series). The core figure unrounded was about 0.33 percent.
- Against consensus. The plan's named consensus was the WSJ survey of 14 forecasters as compiled by investingLive: headline 0.6 percent and 3.6 percent, core 0.22 percent, with a range of 0.16 to 0.24 percent. Headline matched on the month and ran a tenth above on the year. Core was above the top of the range. For how the components fit together, see our guide to CPI components and how CPI differs from PCE.
- Beige Book (2:00 PM ET). Prices rose in every District: three described price growth as robust, eight as moderate and one as modest, and four said the pace had picked up. "Higher oil prices permeated all District reports," with diesel and fuel surcharges named. Activity was weak: two Districts reported moderate growth, three modest, four slight and three no change, and most reported little or no change in employment (source: Federal Reserve, Beige Book, October 2026, prepared at the Philadelphia Fed from information collected on or before October 5). Today's reference piece explains how to read it and compares it with September's.
- Fed. Vice Chair for Supervision Bowman spoke on supervision and regulation in Bangkok at 5:40 AM ET (Federal Reserve calendar). We did not review her remarks and attribute no move to them.
The tape
| Market | Tue Oct 13 reference | Wed Oct 14 (4 PM / close) | Change |
|---|---|---|---|
| 10Y yield | 5.217% | 5.190% | -2.7 bp |
| 30Y yield | 5.563% | 5.525% | -3.8 bp |
| 5Y yield | 5.007% | 4.992% | -1.5 bp |
| DXY (4 PM ET) | 102.198 | 102.521 | +0.32 |
| Gold (GC=F, 4 PM ET) | 4,183.9 | 4,127.3 | -56.6 |
| USD/JPY (4 PM ET) | 158.32 | 158.31 | -0.01 |
| EUR/USD (4 PM ET) | 1.1201 | 1.1159 | -0.0042 |
| Brent (BZ=F, 4 PM ET) | 106.36 | 106.06 | -0.30 |
| S&P 500 | 7,805.45 | 7,718.94 | -1.11% |
Sources: Yahoo Finance query1, the conventions set in the plan: hourly 4:00 PM ET bar for DXY, gold, USD/JPY, EUR/USD and Brent; daily close for ^TNX, ^TYX, ^FVX and ^GSPC. DXY's daily close was 102.55.
The intraday path is the story. The 10-year went from 5.196 percent just before the release to about 5.31 percent by 9:20 AM, then fell for most of the rest of the day, including 3.4 basis points in the half hour that contained the 2:00 PM Beige Book. DXY peaked at about 102.90 in the first hour and gave back more than half of its gain. Gold's low, about $4,101, came around noon. Stocks fell through the session.
Treasury's official curve splits the day this way: the 10-year fell from 5.21 to 5.19 percent, the 10-year TIPS yield from 2.88 to 2.84, and the 10-year breakeven rose from 2.33 to 2.35 (sources: FRED DGS10, DFII10, T10YIE). Inflation compensation went up on a hot print, as it should. Real yields fell by more, and that is what took the nominal yield down.
Grading the plan
The plan had a single leg: September core CPI, seasonally adjusted month on month, as published by BLS to one decimal. Hot at 0.3 percent or more (10-year up at least 3 basis points, DXY up, gold down, USD/JPY up), cool at 0.1 percent or less (the mirror image), no call at 0.2. Graded at the 4:00 PM ET close against Tuesday's reference levels. Classification: 0.3 percent, hot.
| Line in the map | Expected | Actual at close | Grade |
|---|---|---|---|
| Hot: 10Y yield | up 3 bp or more | -2.7 bp (5.217 to 5.190) | Miss |
| Hot: DXY | up | +0.32 (102.198 to 102.521) | Hit |
| Hot: gold | down | -56.6 (4,183.9 to 4,127.3) | Hit |
| Hot: USD/JPY | up | -0.01 (158.32 to 158.31) | Miss |
| Cool (core ≤ 0.1%): all four | mirror image | core 0.3% | Not triggered |
| In line (core 0.2%) | no call | core 0.3% | Not triggered |
| Headline CPI | recorded, not graded | 0.6% m/m, 3.7% y/y | Recorded |
| Beige Book | recorded, not graded | prices up in all 12 Districts | Recorded |
| Brent noise note (record a 3% move before 8:30) | record if triggered | about 105.68 at 8:30 vs 106.36, -0.6% | Not triggered |
The tally is two hits, two misses, no voids. Said plainly:
- The 10-year miss is the one that matters. The plan's bond line was right for an hour and wrong by the close, by about 14 basis points from its peak. A close-vs-levels grade is meant to ignore the first reaction, and it did.
- The USD/JPY miss is a coin toss scored honestly. A change of one hundredth of a yen is no move at all, but the plan said "up," and it was not up. USD/JPY rose 0.43 in the first half hour and gave it all back.
- The dollar and gold hits held up through a day when yields fell, which means they did not hold up on rates. Something else kept the dollar firm and gold soft after the bond move reversed. We do not know what.
Event-window record (not graded). 8:30 to 9:00 AM ET: DXY from 102.20 to 102.75 (+0.55), gold from about 4,175 to about 4,120, USD/JPY from 158.40 to 158.83, and the 10-year from 5.196 to 5.304 percent between the 8:20 and 8:50 bars (+10.8 basis points). From the 8:30 level to the close: DXY +0.33, gold about -48, USD/JPY -0.10, 10-year -0.6 basis points. Everything the bond market did in the event window, it undid by the close. For why we record these windows and do not grade them, see our guide to event windows.
The thesis carry-forward. The asymmetry hypothesis says a hawkish input moves the 10-year up by more than a dovish input moves it down. Wednesday was a hawkish input. In the event window it was strong evidence for the idea (+10.8 basis points). At the close it was evidence against (-2.7). That is now two closes against (October 7 and October 14). The window keeps saying one thing and the close another, so the honest statement is that we cannot yet tell which of the two the hypothesis is about.
The Columbus Day note's two conditional statements
- "If September core CPI prints 0.3 percent or more ... and the 10-year closes Wednesday at or above Thursday October 8's 5.231 percent, then the holiday-week rally in bonds was positioning ... Wrong if a 0.3 percent or higher core print leaves the 10-year below 5.231 at the close." Core printed 0.3 percent; the 10-year closed at 5.190. Graded wrong. On this evidence the holiday-week rally was not just positioning; the bond market kept rallying through a hot print.
- "If the dollar's next move of 0.3 or more in DXY comes with cash Treasury yields moving the same way on the day, then Monday's holiday move was the exception ... Wrong if DXY moves 0.3 or more on a full trading day while the 10-year moves less than 1 basis point or in the opposite direction." Wednesday was that next move: DXY up 0.37 on the daily close (0.32 at 4:00 PM), the 10-year down 2.7 basis points. Graded wrong. The dollar traded apart from rates on a full session, not only on a holiday.
What we do not know
- Why the bond move reversed. Yields fell steadily from 9:20 AM while stocks fell; the S&P 500 lost 1.11 percent. That fits a move into bonds as stocks sold off, and Thursday's overnight session, covered in the next note, points the same way. We have not verified a specific cause for Wednesday's afternoon.
- What the Beige Book did. The 10-year fell 3.4 basis points in the half hour around 2:00 PM, but it was already falling. We cannot separate the report from the trend.
- Fed pricing. We did not read CME FedWatch directly on Wednesday and report no number for it here.
What would change the read
September PPI and advance retail sales followed on Thursday October 15 at 8:30 AM ET; no plan was registered for them, and they are reported in the Thursday note. Because this piece was written after Thursday's close, it adds no conditional statement of its own about Thursday: anything it said would already be known. The forward-looking statements for this thread are in the Thursday note, and a pre-registered map for Friday October 16 was committed to the repository on Thursday evening, before Friday's data.
What to read alongside
- The Beige Book (today's reference piece)
- CPI components
- Breakevens
- Event windows
Nothing on this site is investment advice or a recommendation to trade. Most retail traders lose money. Reaction maps published here are falsifiable hypotheses, graded in public, not signals.