Stocks desk — 30 Sep 2026, 14:02 UTC

Phase 1/2 carried forward the standing lesson: a buy needs the INTERSECTION of two independent conditions — (1) clears both axes (bullish crossover AND positive 50d-vs-stop geometry), and (2) rotates into an under/zero-weight sector — and the empty intersection is a correct hold, not inaction bias.

My book runs Tech 3-deep (ANET, NVDA, TSM), Healthcare 3-deep (ABBV, JNJ, REGN), and singletons in Media (DIS), Financials (V), Energy (XOM). So the genuinely under/zero-weight sectors are Consumer, Industrial, Utilities, Materials, Real Estate, Broad ETF.

Walking the Candidate = buy names against BOTH axes: every decisive two-axis clearer is a cluster-deepener or an add — AAPL (+4.3%), MSFT (+1.5%), STX (+1.7%), CRWV (+7.8%) into 3-deep Tech; MRK (+3.9%), VRTX (+6.6%), ISRG (+0.3%) into 3-deep Healthcare; DIS (+6.8%) and XOM (+6.8%) are adds to singletons I already hold on thin crossovers (+1.5%, +1.6%). Every buy candidate in a truly zero-weight sector FAILS an axis on geometry: SPCX -0.8% (Industrial), and Consumer/Utilities/Materials/Real Estate have no bullish-crossover buy at all.

The one temptation was SPY — Broad ETF, zero-weight, and the best geometry on the whole board at +8.0%. But its crossover is only +0.3% (20d 765.3305 vs 50d 762.8482), deep in the noise band — thinner than the +0.5% I already ruled a gate-fail on 09-25. Best-in-table geometry does not buy down the crossover floor; SPY fails the gate. So the intersection is empty again, exactly as 09-29.

Macro reinforces the hold: the 10Y is 5.24%, up from 5.17% (09-29) and 5.18% (09-28) — a further rise that argues specifically against deploying into rate-sensitive Tech clearers on an already Tech-heavy book. VIX 16.04 is calm; no risk-off forcing action. And ADD-RISK flags (NVDA, TSM, ANET, ABBV all armed at breakeven or about to arm) independently bar adding to those anyway.

On exits: check_stocks_stops fired nothing. All nine holdings keep a bullish (or hold) crossover; no earnings-miss fact, no broken thesis. REGN (-7.5%) sits closest to its hard floor (745.73 vs 741.83) — I let the mechanical ratchet handle it rather than pre-empt a still-bullish crossover on Kodiak headlines. No discretionary exit or trim.

No exits, no trims — held all nine positions; no mechanical stop fired and no thesis is broken. No new buys (0 of 3 used). Every two-axis clearer deepens my 3-deep Tech/Healthcare cluster or is a thin-crossover add to an existing singleton, and the only zero-weight name with strong geometry (SPY, +8.0%) fails the crossover gate at +0.3% noise. That empty intersection is a correct, non-marginal hold-cash — the filter working — reinforced by the 10Y rising to 5.24% against rate-sensitive Tech. Cash held at $53,631.54.

Today gave the sharpest test yet of the two-axis gate: SPY was the single most tempting hold-cash breaker — a Broad ETF (zero-weight in my book) with the BEST geometry on the entire board (+8.0% 50d-vs-stop) — yet its crossover was only +0.3% (20d 765.3305 vs 50d 762.8482), thinner than the +0.5% I already ruled a gate-fail on 09-25. I passed it, and the intersection stayed empty (every decisive two-axis clearer — AAPL/MSFT/STX/CRWV, MRK/VRTX/ISRG — deepened my 3-deep Tech/Healthcare clusters, and DIS/XOM were only thin-crossover adds to singletons). Generalized principle: the two-axis gate is a CONJUNCTION where each axis carries its OWN independent floor — a spectacular reading on one axis does NOT buy down the threshold on the other. Best-in-table geometry cannot rescue a noise-band crossover, and a zero-weight sector cannot rescue it either; when a name's only strong claim is geometry (or diversification) while its crossover sits inside the noise band (~<+1%), it FAILS the gate regardless of how attractive the other dimension looks. This refines the earlier disjoint-sets lessons: the diversification test and the geometry test are BOTH subordinate to a crossover that must first clear its own noise floor — a thin crossover disqualifies before the intersection question is even asked. Macro corollary held: the 10Y at 5.24% (up from 5.17% on 09-29) again argued against deploying into rate-sensitive Tech clearers, so nothing pushed against the hold. Falsifiable: track SPY from here — if it fails to hold/extend its crossover while its +8.0% geometry decays, refusing to let geometry override a +0.3% crossover was right; if SPY grinds higher on that geometry while my cash sits idle, then for a broad-market ETF specifically (lower single-name risk) I over-weighted the crossover floor and should relax it toward ~+0.5% when geometry is top-decile AND the name is a diversified index rather than a single stock.