Stocks desk — 4 Nov 2026, 15:01 UTC

Phase 1 carried two live priors, both on PYPL: a dual-gate name whose trend keeps steepening while the conviction gate stays frozen (11-02) and then moves the wrong way (11-03, sells 3→5). Today PYPL prints bullish again (56.8438 > 55.196) with a buy flag, but its stop geometry is -0.3% (negative — a routine pullback runs through the stop) and its wall is 4/10/33/5/0, still 5 outright sells. Both my priors say the correct read is a firm PASS, not "closer" — geometry alone already rejects it, and the frozen-wall-as-bearish read confirms it.

Mechanical stops fired once: XOM's trail stop (173.42) breached at 171.66 and closed the whole position — it had given back a +15.2% peak to +4.9%, exactly the kind of round-trip the ratchet exists to stop. That frees capital and drops me to 11 positions.

Discretionary pass: nothing to do. Every remaining holding is buy or hold with 20d>50d intact; none flagged sell. UNH is my only red mark (-2.1%) but its trend is bullish (402.2802 > 390.5483) and its wall is strong (8/19/5/0/0) — no broken thesis, so I don't churn it.

New-buy pass: all held buy candidates (AAPL, ABBV, COST, ISRG, ROK, V, DIS, MSFT) are ADD-RISK per the stop check — their ratchets are armed or about to arm, so a normal pullback would stop out the whole position plus any add — so no adds. Among fresh names I used the 50d vs stop column as the gate: PYPL -0.3%, LLY -7.8%, AMGN -3.5%, REGN -10.7%, MRNA -18.1% all reject on negative geometry (a routine pullback to the 50d runs through the initial stop). Only AMZN +2.1% clears it. AMZN is bullish (254.352 > 253.9806), price 270.52 sits above both MAs, analysts are 19/48/3/0/0 (dominant constructive wall, zero sells), earnings are not in the blackout window, and the positive geometry means even a pullback to the 50d holds above the -8% stop. The caution is a thin MA separation (+0.1%) with price extended ~6% over the fast MA — but the protective check (stop sits below the 50d) survives that pullback, so the setup is sound rather than marginal. Sector is Consumer (I also hold TSLA/COST there), acceptable.

Mechanical stop closed XOM (trail 173.42). Held the other 11 — no discretionary exits. Opened one fresh position, AMZN (buy, limit 271.9), the only fresh buy candidate with positive stop geometry (+2.1%), a bullish trend, and a clean analyst wall (19/48/3/0/0). Passed PYPL again (negative geometry -0.3% and a still-frozen 5-sell wall) and rejected LLY/AMGN/REGN/MRNA on negative geometry. 1 of 3 daily buys used.

Bought AMZN as the lone fresh buy with positive stop geometry (+2.1%), rejecting PYPL (-0.3%), AMGN (-3.5%), LLY (-7.8%), REGN (-10.7%), MRNA (-18.1%) on negative geometry. But AMZN's quality is mixed: the trend is bullish only by a hair — 20d 254.352 vs 50d 253.9806, a +0.1% separation — while price 270.52 sits ~6.4% ABOVE the fast MA, an extended hug, not the non-extended +2.5% hug my PYPL lessons treated as healthy. Generalized principle: positive stop geometry and fast≥slow are necessary gates but not a full green light — decompose the trend into (i) MA separation and (ii) price's distance above the fast MA, and read them together. A thin separation WITH an extended price is a weak, not-yet-established trend where price has run ahead of a flat MA structure; the positive geometry only guarantees a pullback to the 50d stays above the hard stop, it says nothing about whether the trend will hold. Prefer a wider separation with a non-extended hug; when forced to choose a thin+extended name, size it as the marginal entry it is, not a conviction buy. This is the mirror of the frozen-conviction-gate idea: there a cleared gate (trend) gave false comfort while another (conviction) was dead; here one trend sub-signal (geometry) gives comfort while the others (separation, extension) are weak. Falsifiable: if AMZN holds above the fast MA and the 20d/50d separation widens, the thin+extended entry was fine and I can relax the extension caution; if instead it mean-reverts toward the flat MAs and round- trips the entry (XOM today gave back a +15.2% peak to +4.9% before its trail fired — the same extended-then-reverted pattern), the extension warning was the real signal and thin- separation+extended-price names should be passed even with positive geometry.