Market Outlook 2026

The stagflation regime broke this month, and it broke on the leg nobody was watching. June CPI printed 3.53% year over year against 4.25% for May, and CPIAUCSL fell 0.42% to 332.568, an outright decline in the index. Labor tightened rather than loosened: initial claims dropped 13.8% to 187,000 on July 18 and unemployment slipped to 4.2%. Financial conditions eased to -0.554 on July 24. The only input still arguing for a growth scare is GDPNow at 1.3%, a July 8 reading that refreshes on Jul 30 next to the Q2 advance print. Falling realized inflation with a tightening labor market is the reflation quadrant, so the trajectory moves to TRANSITIONING with reflation as the target, and the STAGFLATION label survives one more cycle only because the growth evidence is split and Jul 30 settles it. Highest-conviction trade: long the 10-year breakeven against nominals, and conviction across this whole book tops out at moderate. The arithmetic on the entry is what makes it, not the number of ways it wins. Over the matched June 30 to July 28 window DGS10 rose 17bp to 4.61% while DFII10 added 21bp to 2.41%, so the breakeven implied by that pair narrowed 4bp to 2.20%, and T10YIE reads 2.26% on July 29 against 2.24% on June 30. Across almost exactly that stretch the FRED WTI series gained 19.4% to $84.25 and Brent reached $89.27. Two basis points of inflation compensation against a 19.4% move in the crude input is a position, not a neutral price. Where my last version of this argument went wrong was in claiming the trade pays both ways. It does not. A duration-matched long TIPS against nominals is long inflation compensation and nothing else, so falling real yields alone do not pay it: the disinflation branch at 30 narrows breakevens as the realized data cool and kills it, and the growth-crack branch at 20 most likely does the same through demand destruction. Branches one and four pay it, 50 combined, and the honest cover for the other 50 is a separate long at the front end with opposite triggers, not the same position described twice. What the market is getting wrong sits in one place now, and a second candidate has to be withdrawn. The first has partly resolved and the July 21 desk deserves credit: equity vol was called underpriced against credit, and VIX went to 20.66 from 18.65 while HY OAS widened to 2.84% from 2.69% and SPX fell 2.5% to 7,294.6. That divergence is closed and no longer a trade. The candidate I am withdrawing is the front end. T10Y2Y jumped to 0.45 on July 29 from the 0.35 implied by the July 28 DGS10 and DGS2 pair, 10bp in one session, and this pull carries no July 29 print for either leg, so I cannot say whether the 2-year rallied or the 10-year sold off. TNX_10Y at 4.622 on July 29 against DGS10 at 4.61 on July 28 hints at the former, but 1bp across two different sources decomposes nothing. Cut pricing may have started; the evidence here does not establish it, and I carry that as a low-probability thesis rather than a call. Scenario weights: energy-led reflation 35, disinflation holds and the Fed opens the door 30, growth leg finally cracks 20, energy shock and inflation re-acceleration 15. Gold is the only view in the book that improved on its own evidence rather than on a forecast, up 1.16% into a 21bp real-yield rise.