End of week: the thesis missed. Here is the post-mortem.

DXY 101.37 · EUR/USD 1.1383 · USD/JPY 161.74 · US10Y 4.37% · Brent $72.91 · XAU $4083

Two weeks ago, in the original 22 May piece, the call was that the dollar was being moved by the activity side, not by the Fed, and that the dollar's path over the following weeks would be governed by the data flow. The data has now flowed. May nonfarm payrolls printed at +172,000 against an 85,000 consensus this morning. March was revised up by 29,000 to 214,000, and April was revised up by 64,000 to 179,000, for a combined upward revision of 93,000. DXY closed at 101.37, within a quarter point of the round 100 level it has not seen since March. The data-not-dots call did not survive contact with the data. This is the post-mortem.

What we got right

1. The Fed-side framing. Through the whole two-week arc, the OIS curve moved very little. Year-end Fed-funds expectations have firmed by about 10 basis points over the period, and they only moved on the NFP day. The Committee did not reprice in advance; the framing fracture we covered in the Conference Board piece stayed unresolved through Powell's end-of-term statement.
2. The cross-asset architecture. The DXY-Brent regime anomaly we flagged held throughout. The fact that DXY and Brent were co-moving rather than moving inverse was the empirical signature of the geopolitical and growth flow pattern we identified; both unwound together (Brent from $107 to $91, DXY from 99.32 to 98.94) and both rallied together (Brent from $91 to $97, DXY from 98.94 to 99.45) before today's NFP rotation broke the co-movement and restored the textbook DXY-gold inverse.
3. The dollar-smile regime identification. The call was that DXY was sitting in the trough of the smile, where the rate-differential pull was weak and the safe-haven pull was dormant. That description was accurate through the 29 May low. Today's move is the right edge of the smile engaging as the US growth differential against G10 swings meaningfully positive.
4. The discipline of writing through the test. The 1 June piece on ISM, the 2 June piece on JOLTS, and the 3 June piece on ADP each acknowledged the upside surprises in the direction they actually printed, without hedging. That is the editorial discipline this publication runs on. It does not protect a thesis from being wrong; it makes the wrong-ness legible.

What we got wrong

1. The activity-side soft-data persistence. The call was built on the assumption that the early-May activity softening was indicative of a continuing trend. May ISM at 54.0 was the first warning that the trend had reversed; the ADP-plus-NFP sequence confirmed it. The data was not "still soft"; it had turned in early May.
2. The signalling read of the soft 29 May PCE reaction. The PCE piece read DXY's failure to rally on a hawkish PCE print as evidence that positioning was crowded long-dollar and would unwind. The reading of positioning was probably right. The reading of what would follow was wrong. A market that fails to extend on hawkish news can mean crowded-and-tired, but it can also mean priced-in-and- waiting. The week's sequence showed it was the latter.
3. The Beveridge-curve interpretation. The JOLTS piece read the openings-up, hires-down divergence as a frozen labour-market regime. The NFP print and the upward revisions to prior months suggest the hires-fell-while-openings-rose pattern was a one-month anomaly in the JOLTS data, not a structural break in the matching function. The Beveridge framework still works; we over-weighted one observation in it.
4. The translation of the surprise-index move. The 3 June piece said the Citi CESI had not moved enough to flip regime. In fact it had; the index doesn't have to cross a particular threshold for FX to start trading it, and the dollar started trading it the moment ISM surprised. We were one observation late on the regime call.

The two-week arc, in one chart

Where the market sits at end of week

• DXY at 101.37, a quarter point off the 100 line. Up roughly 0.6 percent on the day. Through the 99.5 technical resistance.
• 10-year yield at 4.37 percent, up 6 basis points on the print. The term-premium side stayed firm; the expected-rate piece did the work today. Markets are now pricing roughly a 60 percent probability of a Fed hike by year-end. That is a remarkable number for a Committee that the consensus had pricing 50 basis points of cuts as recently as April.
• EUR/USD broke 1.16 down to 1.1383. The crowded long we flagged on 22 May took the structural hit it was set up for. Spot is now below the 100-day moving average for the first time since March.
• USD/JPY at 161.74, extending higher through the 160 intervention zone. MoF verbal escalation is now at step three of the five-step script; an operation is realistically a one-week probability rather than a one-month probability.
• Gold at $4083, down roughly $100 on the day. The classic dollar-gold inverse is fully back; the correlation cell we flagged as the regime anomaly is resolved.
• Brent at $72.91, back to where it traded before yesterday's geopolitical spike. The premium has bled out; the data does the work.

Through the dollar smile, today

Two weeks ago the US sat in the trough of the smile. Today we are on the right edge. The US growth differential against G10 has swung from modestly negative (the data-not-dots regime) to clearly positive, with the smile's rate-differential pull engaging. The framework still works; the regime call just changed.

The right edge has its own asymmetries. A DXY at 100 sits at a level where MoF intervention probability is meaningfully above zero on the yen leg; where the EUR long position is well below crowded; where the smile's fiscal-risk left-tail Stephen Jen has been writing about becomes relevant if the rally extends. None of those are base cases for next week. All of them are visible asymmetric risks.

The chain so far, end of week two

Twenty published pieces in fourteen calendar days. The arc:

• 22 May. The dollar is weaker on the data, not the dots; EUR positioning has caught up; curve un-inverted without a recession.
• 25 May. Three sessions on; correlations.
• 26 May. A new Chair, a record-low Michigan print; COT positioning.
• 27 May. Confidence beat, bonds rallied; term premium.
• 28 May. DXY finally moved; dollar smile.
• 29 May. Core PCE ticked up; PCE vs CPI.
• 30 May. End of week wrap.
• 1 Jun. ISM came in hot; surprise indices.
• 2 Jun. JOLTS mixed signal; Beveridge curve.
• 3 Jun. Two more upside prints; ADP vs NFP.
• 4 Jun. USD/JPY broke 160; FX intervention.
• 5 Jun. Today's piece; NFP anatomy.

What we are watching next week

• The 16-17 June FOMC. Warsh's first meeting as Chair. The framing he brings to the press conference is the highest-watched moment of the month. The hot NFP removes the case for a near-term cut; the open question is whether the Committee acknowledges the regime change on activity in the SEP and the statement.
• MoF and BoJ on the yen. Verbal escalation is at step three; the next step is the line that means operations are operational. The 24-hour window after a step- four statement is when the price action turns.
• The benchmark-revision risk to recent NFP prints. Today's report carried 93k of upward revisions to the prior two months. The benchmarking pattern of the post-2024 revisions has tended to revise down. The NFP anatomy piece we published this morning walks through the methodology.
• CPI on 11 June. The next inflation print. Through Warsh's stated preference for trimmed-mean PCE, CPI matters less for the Committee than the surprise it generates matters for the CESI.
• The DXY 100 line. The technical resistance sits there. A clean break above on next week's data sets up a quick move to 100.5; a fade from the line keeps the consolidation narrative intact and would be the cleanest reset opportunity for any view that the dollar move is now stretched.

The discipline, restated

The point of writing publicly through a thesis is not to be right. It is to be readable. The data-not-dots call was written explicitly so that this week's outcome could be assessed against it. The thesis was that data, not Fed framing, would drive the dollar. That part was right; the data did the work all week. The thesis was also that the data would continue to soften. That part was wrong. The chain captures both. The next chain starts from the data we have, not from the data we wished for.