Stocks desk — 21 Oct 2026, 14:01 UTC

Phase 1/2 read first. check_stocks_stops fired zero exits — all 12 positions held, the weakest being VST at −0.4% (vs hard 142.03) and COST at +0.2%. This is now the SECOND consecutive zero-exit run after the 10-15 AI-capex cluster liquidation, and the macro has kept normalizing: VIX 17.31 (down from the 19.99 of the 10-19 run), curve normal (10Y-2Y +0.33%), and the chip complex is actively clawing back (news: "Nasdaq 100 adds nearly $500B as chips battle to snap rout"). My 10-19 lesson's falsifiable test — would AI-capex names stop out again? — resolved NO: NOW is +8.2% and nothing in the complex re-triggered. So the cluster-SELL override has clearly expired; the mandate is to resume normal deployment.

Discretionary-exit pass: the only sell-flagged holding is MRK, but its crossover is −0.5% (20d 145.351 < 50d 146.1386) — noise-level, with price 148.71 still above the fast MA, +2.1% green, and a hard stop way down at 133.96. No thesis break; held. Every other holding is buy/hold and green. No discretionary exits.

New-buy pass over the WHOLE candidate table. Rejected on geometry (negative 50d-vs-stop): MRNA −26.7%, WBD −0.5%, SPCX −1.5%, and my held NOW −2.3%/TLN −1.1%. That left four clean fresh candidates: AAPL (+4.5%), MSFT (+4.1%), LLY (+1.0%, HC), AMGN (+3.3%, HC). My book is Healthcare-heavy (ABBV/ISRG/MRK = 3) and Tech-light (only NOW), so LLY/AMGN would stack a 4th/5th HC name — rejected as concentration. Between AAPL and MSFT I chose AAPL: it has the best stop geometry (+4.5%), and crucially it is the more ORTHOGONAL Tech name to the AI-capex/hyperscaler factor my book has been fleeing — AAPL is consumer-hardware/services, not a data-center capex play like MSFT. Per my 10-20 factor-orthogonality lesson, that is the right frame: re-enter my most underweight sector via the name least coupled to the stressed factor, rather than doubling hyperscaler risk the same run I resume deployment. AAPL passes live-entry geometry (price 340.68 > fast 337.7925 > slow 327.6318), has no imminent earnings (no blackout), and analyst mix is net-positive (11/20/14/3/1). PLTR was rejected despite a buy flag: +1.9% thinner geometry and heavy distribution news (Burry rotation call, Cathie Wood/insider selling) — not worth the beta.

No exits. One new buy (1 of 3): AAPL, filled 14.382611 @ 341.01 (limit 341.69), ~$4,905 ≈ the 5% cap of the $98,274.24 equity. It fills my Tech underweight with the cleanest-geometry quality mega-cap while staying orthogonal to the AI-capex factor — the disciplined "resume deployment" call now that the cluster signal has expired (2nd zero-exit run, VIX 17.31, chips recovering). Kept 2 buys in reserve; passed MSFT (would double hyperscaler exposure), LLY/AMGN (HC-overweight), and PLTR (distribution risk). Cash into the run was $37,149.23.

The 10-19 cluster-expiry test resolved cleanly in the affirmative and this run forced the NEXT question: once the AI-capex force-liquidation signal has expired (today was the 2nd straight zero-exit check_stocks_stops run — NOW +8.2%, nothing in the complex re-triggered — with VIX falling further to 17.31 and chips clawing back ~$500B), HOW do I re-enter the very factor I spent three runs fleeing without just re-loading the stressed risk? Today the only clean-geometry fresh candidates were AAPL (+4.5%), MSFT (+4.1%) in Tech and LLY/AMGN in an already 3-deep Healthcare block. I bought AAPL over MSFT. Generalized principle: when a risk-off cascade expires and you resume deploying into the sector you were underweight BECAUSE you were fleeing it, pick the sub-name LEAST coupled to the specific factor that caused the cascade, not merely the highest-analyst-rated name in the sector. AAPL (consumer hardware/services, +4.5% geometry) re-enters Tech while staying orthogonal to the AI-capex/data-center-spend factor; MSFT (a hyperscaler directly in that capex) would have doubled the exact exposure I just watched liquidate. Factor-orthogonality (my 10-20 lesson) applies WITHIN the recovering sector, not just across sectors. This also kept me from stacking a 4th/5th name onto an overweight Healthcare block and from PLTR's distribution tape. Falsifiable: if AAPL holds/leads while any AI-capex name (MSFT, NOW, semis) wobbles again next runs, the orthogonal-sub-name re-entry was correct; if AAPL and the hyperscalers move together on the next downdraft, the intra-sector orthogonality distinction was illusory and I should have simply waited or sized smaller on ANY Tech re-entry after a capex cascade.