Wed: hawkish minutes, a strong 10-year auction, and yields that went nowhere. Plan graded.
Covers Wednesday October 7, with the pre-registered plan graded at the close. The September FOMC minutes said most participants saw another hike as likely appropriate by year end: hawkish by the plan's rule. The $39 billion 10-year reopening sold at 5.300 percent, a 1.7 basis point stop-through, with a 2.77 bid-to-cover and 80.3 percent indirects (Treasury Fiscal Data): strong. The two legs pointed the 10-year in opposite directions, so its call is void; it closed up 0.8 basis points at 5.277 percent. DXY up and gold down were hits, USD/JPY a two-pip miss. Candidly, nearly all of the dollar and gold moves came before the minutes were released.
Covers Wednesday October 7, written after the close and finished early Thursday October 8, before any Thursday US data. A morning plan was committed to the repository on Tuesday evening, before Wednesday's releases, and it is graded below. Wednesday had two scheduled tests for the rates market, and both came in on the side the plan defined in advance. The minutes of the September 15-16 FOMC meeting said most participants expected another rate increase to be appropriate by year end: hawkish by our definition. The $39 billion 10-year reopening drew a 2.77 bid-to-cover and an 80.3 percent indirect share: strong by our definition. Those two outcomes pointed the 10-year in opposite directions, and the 10-year finished up 0.8 basis points, at 5.277 percent. The dollar rose and gold fell, as the hawkish leg said they would, but almost all of both moves happened before the minutes came out.
This continues Tuesday's note on ISM Services prices and the 10-year at 5.31, which ended with two conditional statements about Wednesday. Both are scored below as well.
What the minutes said
The minutes were released at 2:00 PM ET (source: Federal Reserve, Minutes of the FOMC, September 15-16, 2026). The September meeting raised the target range 25 basis points to 3.75 to 4.00 percent on a 12-0 vote. The sentence our plan was built to find is in the policy section:
"With regard to the outlook for monetary policy beyond the current meeting, most participants assessed that another increase in the target range for the federal funds rate would likely be appropriate by year end."
Other lines pointed the same way. "Many participants emphasized that a higher path for the target range would be prudent on risk-management grounds." "A number of participants viewed a higher path for the target range as necessary based on their modal outlooks." "Several participants stated that they viewed the current policy rate as not restrictive or only mildly restrictive," and "a couple of participants" said they had raised their estimate of the neutral rate. Participants "generally assessed inflation risk as skewed to the upside." We found no passage in which most or many participants judged it appropriate to hold the rate for a time. For how the Fed's quantifier words rank ("a few," "some," "several," "many," "most"), see our guide to reading FOMC minutes.
Two details worth recording for later. The staff noted that "changes in real rates contributed to most of the net increase in longer-maturity Treasury yields" over the intermeeting period. And the Desk's market review cited "competition for capital from heavy private debt issuance to finance the development of artificial intelligence (AI) infrastructure" as one factor in higher term premiums. Both match what Monday's official data showed: a yield rise carried by real yields, not breakevens.
What the 10-year auction showed
Treasury sold $39 billion of the 10-year note (CUSIP 91282CRF0, a reopening of the August issue) at a high yield of 5.300 percent. Bid-to-cover was 2.77, up from 2.71 at the September reopening. Indirect bidders took 80.3 percent of the competitive amount accepted ($31.06 billion of $38.67 billion), up from 79.2 percent; direct bidders 17.1 percent; primary dealers 2.5 percent (source: Treasury Fiscal Data, auctions query). The when-issued yield at 1:00 PM was 5.317 percent, so the auction stopped through by 1.7 basis points (sources: investingLive, Newsquawk; Treasury does not publish the WI level). Our guide to auction demand treats a stop-through of 1 basis point or more as materially strong. This one was.
The cash 10-year traded at about 5.32 percent just before 1:00 PM and about 5.28 percent by 1:20 PM (Yahoo ^TNX 30-minute bars). That 4 basis point drop in the auction window was the largest single move of the session.
The rest of the calendar
- EIA petroleum (10:30 AM ET), week ending October 2. Commercial crude inventories fell 3.2 million barrels to 424.1 million; total motor gasoline rose 0.4 million to 204.7 million; distillate was flat at 105.1 million; the Strategic Petroleum Reserve fell 0.8 million to 283.0 million (source: EIA Weekly Petroleum Status Report, Table 1). Brent futures traded between about $99.6 and $102.6 on the day and finished near $100.95 at 4:00 PM ET, down about $0.20 (Yahoo BZ=F hourly bars). We do not attribute the intraday swing to the EIA numbers.
- Consumer credit (3:00 PM ET), August. Up at a 1.9 percent annual rate, after 4.1 percent in July. Revolving credit fell at a 4.2 percent rate; nonrevolving rose at 4.1 percent (source: Federal Reserve G.19). The plan made no call on it, and we see no market reaction to report.
The tape
| Market | Tue Oct 6 reference | Wed Oct 7 close | Change |
|---|---|---|---|
| 10Y yield | 5.269% | 5.277% | +0.8 bp |
| 30Y yield | 5.641% | 5.661% | +2.0 bp |
| 5Y yield | 5.028% | 5.021% | -0.7 bp |
| DXY | 101.83 | 102.25 | +0.42 |
| Gold (GC=F, 4 PM ET) | ~4,193 | ~4,137 | ~-56 |
| USD/JPY (4 PM ET) | 158.07 | 158.05 | -0.02 |
| EUR/USD (4 PM ET) | 1.1264 | 1.1201 | -0.0063 |
| Brent (BZ=F, 4 PM ET) | ~101.15 | ~100.95 | ~-0.20 |
| S&P 500 | 7,818.93 | 7,801.77 | -0.22% |
Sources: Yahoo Finance query1 daily closes for ^TNX, ^TYX, ^FVX, DX-Y.NYB and ^GSPC; Yahoo hourly 4:00 PM ET bar closes for GC=F, USDJPY=X, EURUSD=X and BZ=F, the same convention as the reference levels in the plan.
The shape of the day matters more than the closing numbers. The 10-year opened near 5.34 percent and touched about 5.36 in the first hour, then drifted down through the morning, dropped on the auction, and was within a basis point of its 2:00 PM level at the close. DXY did most of its rising overnight: it was about 102.4 by 7:30 AM ET and 102.29 at 12:45 PM. Gold fell from about 4,148 to about 4,106 in the half hour from 8:00 AM ET and briefly traded below 4,100, then spent the afternoon between 4,127 and 4,146. From 2:00 PM to 4:00 PM ET, DXY moved from 102.24 to 102.28 and gold from about 4,136 to about 4,137 (Yahoo 15- and 60-minute bars). In other words, the minutes did not move the dollar or gold much at all.
Grading this morning's plan
The plan, committed Tuesday at about 9:10 PM ET, had two legs and a conflict rule. Leg A classified the minutes by their own quantifier words: hawkish if "most" or "many" participants judged further increases likely appropriate, which called for the 10-year up at least 4 basis points, DXY up, gold down and USD/JPY up. Leg B classified the auction: strong if bid-to-cover was at least 2.70 and indirects at least 78 percent, which called for the 10-year to close at or below Tuesday's 5.269 percent. The conflict rule: if the two legs pointed the 10-year in opposite directions, the 10-year call is void for both. Everything is graded at the 4:00 PM ET close against Tuesday's reference levels.
Classification: Leg A is hawkish ("most participants assessed that another increase ... would likely be appropriate by year end"; the dovish definition is not met). Leg B is strong (2.77 and 80.3 percent, both above the bars). Hawkish says the 10-year goes up; strong says it does not. The conflict rule fires.
| Line in the map | Expected | Actual at close | Grade |
|---|---|---|---|
| Leg A: 10Y yield | up 4 bp or more | +0.8 bp (5.269 to 5.277) | Void (conflict rule) |
| Leg A: DXY | up | up 0.42 (101.83 to 102.25) | Hit |
| Leg A: gold | down | down ~56 (~4,193 to ~4,137) | Hit |
| Leg A: USD/JPY | up | down 0.02 (158.07 to 158.05) | Miss |
| Leg B: 10Y yield | at or below 5.269% | 5.277% | Void (conflict rule) |
| Mixed / Dovish (Leg A), Weak / Neither (Leg B) | not met | Not triggered | |
| Brent noise note (over 3% intraday on a headline before 2 PM) | record if it happens | range about 2.9%, no headline attributed | Not triggered |
| EIA petroleum, consumer credit | no call | Not graded |
The tally is two hits, one miss and one void. Three things to say about it, plainly.
- The void is not a rescue. Had either leg stood alone, the 10-year call would have missed: +0.8 basis points is short of Leg A's 4, and 5.277 is above Leg B's 5.269. The conflict rule was written before the data, and we apply it as written, but the reader should know the 10-year line would have failed on both counts.
- The two hits are weaker than they look. The close-versus-levels rule credits any move between Tuesday's close and Wednesday's close. Almost all of DXY's rise and all of gold's fall happened before 2:00 PM, much of it before New York opened. The plan got the direction right; there is little evidence the minutes caused it. That is a flaw in how the map was graded, not a reason to change this grade. Thursday's plan adds an event-window record (not graded) so this gap is visible in future write-ups. The reference piece published alongside this note explains the method.
- USD/JPY missed by two pips. It is a miss. The pair rose about 0.10 in the two hours after the minutes, the direction the map called, but the close is the rule.
The thesis carry-forward. The plan kept open the asymmetry hypothesis: a hawkish input moves the 10-year up by more than a dovish input moves it down. Wednesday gave a clearly hawkish input, and the 10-year rose 0.8 basis points on the day and was roughly flat from 2:00 PM to the close. That does not support the hypothesis. It is not a clean test either: the minutes were largely priced. FXStreet, citing CME FedWatch, put the chance of at least one more hike by year end at about 85 percent before the release (source: FXStreet, Oct 7), and the minutes predate the September payrolls report. A hawkish document that confirms the price is a smaller input than a hawkish data surprise. Recorded as one observation against, not a refutation.
Tuesday's two conditional statements
- "If the minutes describe most or many participants seeing further increases as likely appropriate and the 10-year rises through Monday's 5.311 percent close ... Wrong if a hawkish minutes day ... fails to lift the 10-year at all." The first half was met; the 10-year did not get near 5.311 at the close and rose less than a basis point. Graded wrong. The breakeven half cannot be checked yet: Treasury's official October 7 yields were not on FRED at writing (October 6 was: 10-year 5.27, 10-year TIPS 2.91, breakeven 2.36, the same as the prior two sessions; sources: FRED DGS10, FRED DFII10).
- "If the 10-year reopening shows a lower indirect share than September's 79.2 percent and bid-to-cover below September's 2.71, then Tuesday's softer 3-year demand was not a one-off. Wrong if the 10-year sale comes in at or above both September figures." It came in above both (80.3 percent, 2.77). Graded wrong. On this evidence, Tuesday's soft 3-year looks like a one-off.
What we do not know
- Why the 10-year rose about 7 basis points overnight before falling back. Pre-market reports pointed to inflation worries, deficits and AI-related debt issuance (FXStreet), all of which are slow-moving and none of which changed on Wednesday. We cannot separate them.
- Who the indirect bidders were. "Indirect" covers foreign official and private accounts and domestic funds bidding through dealers. An 80 percent share says demand was broad; it does not say it was foreign.
- Current Fed pricing. We have not captured a post-minutes reading directly from CME and do not report one.
What would change the read
Thursday brings initial jobless claims at 8:30 AM ET and a $22 billion 30-year reopening at 1:00 PM ET (source: Treasury Fiscal Data). A conditional map for both was committed to the repository before the data and will be graded at the Thursday close. No unconditional call here; two conditional statements:
- If the 30-year reopening draws an indirect share of at least 77 percent and a bid-to-cover of at least 2.60 (September: 79.5 percent and 2.61, per Treasury Fiscal Data), then Wednesday's 10-year result was part of a run of strong long-end demand, not a one-day event. Wrong if either figure falls short.
- If Treasury's official October 7 curve shows the 10-year breakeven unchanged at about 2.36 again, then a fourth straight session has kept inflation compensation flat while nominal yields moved, consistent with the real-yield reading in the minutes' own staff review. Wrong if the breakeven moved by 3 basis points or more.
What to read alongside
- Event windows: how to tell what a catalyst actually moved (today's reference piece)
- How to read FOMC minutes
- Treasury auction demand
- Friday recap: our last graded plan
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.