Saturday, 1 August 2026

Market regime

Risk-on — as of the last close.

The week in review

The model's latest read is risk-on, its first risk-on print, a flip from neutral. That flip rests on three measured inputs, not on a narrative: SPY closed at 746.79 against a 200-EMA of 700.92, putting the index +6.5% above trend; breadth sits at 59.6% of scanned US stocks above their 50-day average (581 of 975), which the model classes as healthy but marginal; VIX at 15.99 sits in its calm band. None of those is dissonant, though breadth is the thinnest of the three. A tape 6.5% above trend with participation this broad sits comfortably with a calm VIX, and the model reads risk-on because all three inputs clear.

Rates did the work under the surface. The 2Y closed at 4.23% as of 30 July 2026, down 10bps on the week; the 10Y at 4.68% moved 1bp lower over the same span. The 10Y-2Y spread printed 0.47% as of 31 July 2026, +11bps week-over-week the front end moved and the long end did not follow. Breakevens (T10YIE) at 2.28% ticked +2bps; the real 10Y at 2.40% eased 3bps. This is a front-end repricing, not an inflation-expectations story: the breakeven barely moved while the 2Y did all the traveling.

Credit disagreed, mildly. HY spreads (BAMLH0A0HYM2) widened +5bps to 2.84% as of 30 July 2026. That is a small move off a tight base, and it is the only weekly series pointing the other way from equities. Initial claims at 197K (week of 25 July 2026) rose 9K week-over-week still a low absolute level, but the direction is worth logging rather than dismissing. Older monthlies frame the backdrop: unemployment 4.2%, payrolls 159.0M, Fed Funds 3.63%, housing starts 1,427K, all as of 1 June 2026 and all stale relative to the week's action.

Measured: the regime flip, the level of every figure above. Inferred: that a bull-steepening front-end move plus widening HY spreads is a cut-expectations trade rather than a growth-acceleration trade. That inference is falsifiable see The Week Ahead.

Themes in motion

Six themes carry an accelerating tag, and the composition is the tell: AI datacenter infrastructure (14 names), oil, energy & geopolitical (11), medtech & diagnostics (11), M&A & special situations (10), AI chips & memory (10), AI enterprise software (10). AI now occupies three of the six accelerating slots and 34 tracked names across them.

Four themes have rolled to maturing: industrial power & grid (12 names), mega-cap AI platforms (8), fintech & consumer credit (6), rare disease & gene therapy (6). Mega-cap AI platforms maturing while AI chips, datacenter infrastructure and enterprise software all accelerate is the week's most legible rotation breadth moving down-cap within the same secular story. Industrial power & grid maturing alongside accelerating datacenter buildout is the harder one to reconcile; the picks-and-shovels layer for compute was supposed to lead the power layer, not decouple from it.

Fintech & consumer credit maturing is worth pairing with the +5bps HY widening and the +9K claims print. Three weak signals, none conclusive on its own, all pointing at the consumer.

Under the lens

342 dossiers were deep-refreshed this week. The named slice runs alphabetically from AAOI through AMN, which means the refresh is a systematic sweep rather than a conviction-weighted selection coverage, not a shortlist. Reading it as "these are the ideas" would be a category error.

Within that slice, the names that sit at theme intersections:

• AMAT (Applied Materials), ACLS (Axcelis), ACMR (ACM Research), AEHR (Aehr Test Systems), AMKR (Amkor) semicap and test/packaging, the layer beneath AI chips & memory. Five of them in one alphabetical block is a density signal about how the theme is populated, not about relative quality.
• ALAB (Astera Labs), AAOI (Applied Optoelectronics), AIP (Arteris) connectivity and interconnect IP, the AI datacenter infrastructure theme's least mega-cap-correlated corner.
• AMD (Advanced Micro Devices) and AAPL (Apple) the mega-cap AI platform layer now tagged maturing.
• AGX (Argan) and AAON industrial power & grid and HVAC, also maturing despite the datacenter acceleration above them.
• ABT (Abbott), ABCL (AbCellera), ABVX (Abivax), AGIO (Agios), AKBA (Akebia), ALMS (Alumis) medtech & diagnostics (accelerating) and rare disease & gene therapy (maturing) side by side.
• AFRM (Affirm) the fintech & consumer credit read, against a maturing theme tag and a widening HY tape.

Highest-conviction reads are constrained by what the data actually supports. Two survive: (1) the semicap/interconnect cluster is where an accelerating theme has the most tracked names and the least mega-cap beta, and (2) the mega-cap-maturing / down-cap-accelerating split is a real rotation in the theme tags, observable this week. Both are theme-level observations. No single-name conviction is supportable from a coverage sweep a 342-dossier alphabetical refresh tells you what was read, not what was concluded.

The week ahead

The calendar is dense and front-loaded. ARCB and ATAI print at T-0d (1 August 2026). USO carries a 2 August 2026 catalyst at T-1d. Then a cluster of twelve on 3 August 2026 (T-2d): ADTN, AEIS, AESI, BRBR, CCJ, CODI, COIN, EPC, GBTG, HOOD, LIND, MMYT.

Three of those map onto live theme reads and are the ones worth watching:

• AEIS and ADTN power conversion and network infrastructure, both dossier-refreshed this week, both reporting into the datacenter-accelerating / grid-maturing contradiction flagged above. If the grid layer is genuinely decoupling from compute demand, these prints are where it shows up first in guidance language.
• COIN and HOOD on the same day a paired read on retail risk appetite. They report into a maturing fintech theme, a calm VIX, and HY spreads that widened. If the risk-on regime read is real at the retail level, two same-day prints are an unusually clean test of it.
• CCJ and AESI uranium and energy services, into the accelerating oil/energy/geopolitical theme.

The view and what breaks it. The inference above is that this week's move was a front-end rate repricing, not a growth re-acceleration. Invalidation: if next week's data pulls the 2Y back toward the 10Y and the 10Y-2Y spread gives back its +11bps, the bull-steepener read is wrong and the move was positioning, not policy expectations.

Second view. The model's risk-on call is carrying healthy-but-marginal breadth (59.6%) as its thinnest input. Invalidation: breadth falling out of its healthy-but-marginal band from 59.6%, or HY spreads extending beyond this week's +5bps widening off 2.84%, would mean the inputs that carried the upgrade have stopped carrying it and the flip from neutral was early. Conversely, breadth climbing back into its healthy band while VIX holds in its calm band would confirm the model's read.

What is not knowable from here. Whether the +9K claims uptick to 197K is noise or an inflection. One weekly print off a low base supports no conclusion; the series needs several more weeks before it means anything. The monthly labor data in hand (4.2% unemployment, 159.0M payrolls) dates to 1 June 2026 and cannot arbitrate it.