The Overnight Setup

Markets are holding a bull-trending posture with all four majors above their moving averages and volatility calm — but the whole structure leans on one number: the 30-year yield. Everything else today, from sector leadership to a batch of congressional filings, reads as either confirmation of that setup or a live test of it.

The Regime, in One Sentence

Soft headline CPI (3.7%, still sticky) knocked the dollar down and steepened the curve, and that steepening is doing real work: financials (XLF, +7% over one month) and industrials are picking up the baton from AI-capex-driven tech (XLK, +26% over three months), broadening the rally beyond mega-cap names. VIX at 16.5 and in contango confirms no stress in the system. The constraint is the long end — 30-year yields at 5.09% are capping bond-proxies, staples, small caps, and rate-sensitive real estate and materials, and the entire tilt is built around avoiding that basket. Confidence in this read is 78, not 95 — a lean, not a certainty, and it should be sized that way.

Overnight: Noise, Not Signal

The only real 1-hour movement is in speculative alt-crypto names — GWEI, ALLO, KITE, AMP, and a handful of others, each moving 1–4% with no shared theme. That's froth, not a catalyst, and it doesn't touch the equity thesis one way or the other.

The Risk Hierarchy — Ranked, Not Tied

1. The 30-year auction and yield path is the fulcrum. A push toward 5.25%+ starts to bite growth multiples and small caps directly, and the entire "quality large-cap with high-beta kicker" tilt depends on the long end staying range-bound. This is the number that would flip the brief.
2. AI-capex commentary is the second-order risk — smaller in probability but fast-moving if it hits. The melt-up in tech leadership is thin enough that a wobble in hyperscaler spending guidance could hit XLK and momentum names quickly.
3. Bank earnings cadence — NII guidance and capital-markets commentary either confirm the financials breadth leg or stall it.
4. Dollar follow-through — soft-CPI dollar weakness is a tailwind for multinational earnings translation; a snapback would pressure the tech leadership still carrying the tape.

Congressional Filings: Mostly Confirmation, One Real Dissent

A cluster of filings lines up with the AI-infrastructure overweight already embedded in the regime: buys in APH and COHR (connectivity and photonics — the physical layer under AI infrastructure, less crowded than the mega-cap names) and MU (memory and DRAM, an AI-capex derivative via HBM demand). These filings don't predict the regime — they simply happen to sit inside it, which is worth noting and nothing more.

IBM fits the same logic from a different angle: repeat buying reads as a value-oriented way into the AI trade through consulting and software monetization rather than paying up for XLK multiples. It confirms the thesis while offering a cheaper entry point — a variant, not a contrarian signal.

HD is the one filing that cuts against the regime. The buying pattern reads as a curve-steepening, housing-recovery bet — but the regime's entire logic is built around avoiding rate-sensitive exposure while the 30-year sits above 5%. If housing-sensitive names start working despite high long rates, that's a genuine dissent signal worth taking seriously. Until then, it's a watchlist item, not a thesis-confirming one.

Separately, a healthcare cluster — UNH, MCK, ABT — sits outside the current sector tilt entirely. UNH reads as a post-drawdown value bet, MCK as defensive earnings visibility on GLP-1 volume, and ABT shows the most bidirectional activity of the group, more likely reimbursement or tariff-policy noise than directional conviction. None of this extends the equity thesis into healthcare, and it shouldn't be read as if it does.

So What

Nothing overnight moves the needle — the crypto chop is froth, not information. The real decision points today: don't add rate-sensitive exposure (small caps, staples, real estate) while the 30-year sits above 5%, full stop. APH, COHR, and MU are a legitimate way to extend the AI trade without paying mega-cap multiples — worth a closer look, not a chase. And HD goes on a watchlist, not a buy list: it's only interesting if the 30-year move breaks the other way, at which point it stops being noise and starts being the tell that the regime is cracking.

The takeaway: this is a bull market with one lever — watch the 30-year before anything else on the list.