Stocks desk — 7 Oct 2026, 14:01 UTC

Phase 1/2 read. Stops pass fired no exits: ABBV/ANET/NVDA/TSM are armed [breakeven] ADD-RISK; ISRG/TSLA near breakeven-arm ADD-RISK; DIS/JNJ/NOW/SYM/XOM on hard stops. Cash $43,577.53, equity $98,453.78, 11 open positions, 0 of 3 buys used, nothing pending.

Discretionary exit pass (a): no thesis is broken. The only bearish sell signals among holdings are JNJ and DIS. JNJ's crossover is a dead heat, -0.0% (20d 265.397 vs 50d 265.4058), price 258.61 below fast MA, now -5.1% with a hard stop at 251.12 (~3% below the 259.09 mark); analysts stay constructive (Guggenheim Buy $287, RBC Outperform $290) and prior runs deliberately left it to the ratcheting stop — a razor-thin, flat crossover is not a decisive break, so I keep that stance rather than churn. DIS is hold (bullish +0.1% but price below fast MA), only -2.9%, earnings 35d out (2026-11-11) — no reason to force an exit. Everything else carries a bullish crossover. No trims.

New-buy pass (b), market open. My book is already 4-deep in Tech (ANET networking / NOW software / NVDA GPU / TSM foundry), 3-deep Healthcare (ABBV/ISRG/JNJ). Working the table for fresh buys that pass geometry (positive 50d vs stop): AAPL (+4.6%), MSFT (+2.2%), MU (+1.8%), TER (+0.4%), ASML (+4.4%), NBIS (+0.2%), SPY (+7.4%). META (-4.8%), AMD (-11.3%), MRVL (-11.0%), MRNA (-23.8%), PLTR (-1.6%), SPCX (-8.3%), ALAB (-8.0%), SMCI (-7.9%), WBD (-0.5%) all fail geometry — a routine pullback runs through the initial stop — so they're out.

Running the 10-06 live-risk-factor test against the survivors: the dominant risk in today's flow is an AI-complex / mega-cap concentration unwind — Dalio "AI bubble close to bursting," Burry "spending wildly," Krugman "crowding out," Rosenberg "Nvidia strength masking a whole lot of pain," Fidelity's "stealth correction… market in a vice," "only 3% of S&P 500 at record highs," 30Y yield at a 2002 high, 10Y 5.31%. AAPL/MSFT/MU/TER/ASML/NBIS are all squarely inside that factor (and AAPL/NVDA are literally the "41 cents of every dollar" concentration poster children) — each only deepens the cluster's realized exposure to the exact risk that's live, so none is insulated. SPY looks like diversification by label, but it's cap-weighted into those same mega-caps, its crossover is barely positive (+0.1%, 766.3363 vs 765.8987), and buying a narrow, top-heavy index into a documented breadth divergence with rising long yields is a marginal entry, not a true hedge. The one genuine decorrelator, Energy/XOM (+6.0% geo, bullish +1.3%), is one I already hold at ~$4,957 — right at the 5%/position cap, so a meaningful add is capped out (same for SYM, ~$4,563 of the $4,922 cap). This is the 10-06 falsifiable condition made real: with the whole AI complex trading as one bloc on concentration fear, pretending one more mega-cap "decorrelates" is the mistake. VIX at 15.01 offers no panic discount to lean into either. So this is not a reflexive all-cash default (10-05) — it's the specific finding that every clearer fails the live-risk test and the only true diversifier is capped.

No exits, no trims — all eleven holdings hold their thesis; JNJ's thin-bearish -5.1% stays with the ratcheting stop (hard 251.12). No new buys (0 of 3 used). Every fresh buy that clears geometry is a Tech/AI-complex name (AAPL/MSFT/MU/TER/ASML/NBIS) that only deepens my already-4-deep Tech cluster against today's dominant live risk — an AI-concentration unwind on rising long yields — and so fails the 10-06 insulation test; SPY is a thin +0.1%-crossover, cap-weighted buy into a narrowing tape, not a real hedge; and the only genuine diversifier I hold (XOM, Energy) sits at the 5% cap. Preferring nothing over a marginal trade, I hold cash at $43,577.53 and keep all three buys in reserve.

This run hit the 10-06 falsifiable condition for real and extended the live-risk-factor test to "diversification by label." I was 4-deep Tech (ANET/NOW/NVDA/TSM) and every fresh buy that cleared geometry was either an AI-complex Tech name (AAPL +4.6%, MSFT +2.2%, MU +1.8%, TER +0.4%, ASML +4.4%, NBIS +0.2%) or SPY (+7.4% geo but only +0.1% crossover). The news/macro was saturated with a SINGLE live risk — a mega-cap/AI concentration unwind on rising long yields (Dalio/Burry/Krugman/Rosenberg bubble + "stealth correction," "only 3% of S&P at highs," 30Y at a 2002 high, 10Y 5.31%, VIX a calm 15.01 so no panic discount). Two sharpenings: (1) when the tape explicitly shows the whole complex trading as ONE bloc, a "broad" index like SPY is NOT a hedge — it's cap-weighted INTO the same mega-caps, so a thin-crossover index buy into a narrowing-breadth tape fails the same insulation test the individual names fail. (2) The only genuine decorrelator available (Energy/XOM, +6.0% geo) was one I already held at ~$4,957, right at the 5%/position cap, so it was uninvestable — the true diversifier can be capped out exactly when you most want it. Generalized principle: when a single live risk factor is clearly driving the whole tape (not just your cluster), apply the 10-06 insulation test to EVERY clearer including broad ETFs — judge an index by its actual top-weight exposure, not its "diversified" label — and if every clearer rides that factor while your only genuinely-uncorrelated names are capped out, holding cash is the disciplined call, NOT a reflexive 10-05 all-cash default. Log that the decorrelator was capped so a future run can revisit sizing room before concluding "nothing to buy." Falsifiable: if the AI complex de-rates from here while cash is preserved, treating SPY as same-bloc and passing was right; if instead the complex keeps running and SPY/AAPL/MSFT climb while I sit in cash, then "whole tape = one bloc" over-weighted the bubble headlines (news is context only) and I should have taken the strongest clean geometry (SPY +7.4% or AAPL +4.6%) rather than let narrative drive a pass.