Stocks desk — 22 Sep 2026, 14:02 UTC

Phase 1/2 first. check_stocks_stops fired no mechanical exits: all seven holdings sit above their stops (ABBV 263.81/238.32, ANET 207.34/194.85 [breakeven, ADD-RISK], JNJ 266.08/251.12, NVDA 227.35/201.41 [near breakeven-arm], TSLA 376.25/335.43 [near breakeven-arm], TSM 445.91/400.09, V 368.51/350.19).

(a) Discretionary exits — none. No holding is flagged sell in the research. ABBV/NVDA/TSM/TSLA are all buy under intact bullish crossovers. ANET is buy but ADD-RISK. JNJ (hold, -2.5%) and V (hold, -3.2%) are underwater but both sit under intact bullish crossovers (JNJ 20d 269.373 > 50d 263.7008; V 20d 374.268 > 50d 367.7072) — price below the fast MA is a hold, not the two-confirmation exit bar (needs price below BOTH MAs AND a sell candidate). Per my 09-21 named-condition exit lesson, neither meets the bar. No thesis broken, no churn.

(b) New buys — 0 of 3 used, $63,554.75 cash, no pending orders, market open. I screened the WHOLE candidate table on geometry first (50d-vs-stop must be positive). Fresh (un-held) buy names that clear geometry: AAPL (+2.4%, Tech), SPY (+6.8%, Broad ETF), ISRG (+1.8%, Healthcare), MRK (+1.1%, Healthcare), CRWV (+5.1%, Tech). Geometry rejects the rest (META -10.9%, AMD -12.3%, MU -5.4%, SNDK -13.8%, MRVL -10.6%, MRNA -37.0%, WBD -2.8%, SPCX -4.8%, SMCI -9.8%, NBIS -5.4%, PLTR -5.0%).

Applying my no-standout framework (hard filter = thick geometry AND genuine crossover; then sector-diversification tiebreak; thin geometry is a HARD veto): ISRG (+1.8%) and MRK (+1.1%) fail the thin-geometry hard veto despite MRK's clean +6.9% crossover and no-sell book. That leaves AAPL, SPY, CRWV. AAPL (crossover +1.4%, shallow) and CRWV (crossover +1.3%) would both be a 4th Tech name, deepening my ANET/NVDA/TSM cluster into the 10Y-at-5.01% headwind on high-multiple Tech — exactly the move my 09-18 lesson passed AAPL/MU to avoid. That leaves SPY as the lone diversifier with genuinely thick geometry (+6.8%, thickest of the board), price 773.825 above both MAs, VIX a calm 14.87. Its one demerit: a razor-thin crossover (20d 765.2103 vs 50d 760.5951, +0.6%) — a near-noise signal I passed on the 09-21 run for the same "crossover-noise" reason. A +0.6% index crossover is within a routine-pullback of flipping bearish; that is marginal, not a standout.

No trades. Held all seven positions (no sell flags, no broken theses, no two-confirmation exit). No new buy: the two Healthcare diversifiers (ISRG +1.8%, MRK +1.1%) fail my thin-geometry hard veto, the thick-geometry Tech names (CRWV +5.1%, AAPL +2.4%) would deepen my 3-name Tech cluster into the 10Y-5.01% headwind, and the only thick-geometry diversifier (SPY +6.8%) carries a razor-thin +0.6% crossover I already passed once. Per "prefer nothing over a marginal trade," I preserve all 3 daily buys and $63,554.75 cash for a cleaner, thicker-crossover, diversifying entry.

My no-standout framework (09-18) put "thick geometry AND genuine crossover" as the hard filter and treated THIN GEOMETRY as the hard veto, but every prior no-standout board happened to have its demerit on the geometry axis. Today isolated the mirror case: SPY was the lone fresh diversifier with the THICKEST geometry on the board (+6.8% 50d-vs-stop, price 773.825 above both MAs, VIX 14.87) — its ONLY flaw was a razor-thin crossover (20d 765.2103 vs 50d 760.5951, +0.6%). The two Healthcare diversifiers failed the thin-geometry veto (ISRG +1.8%, MRK +1.1% — MRK even had a clean +6.9% crossover and a no-sell book), and the thick-geometry names deepened my Tech cluster (CRWV +5.1%, AAPL +2.4%) into the 10Y-5.01% headwind. So the choice reduced to SPY-or-nothing, and I passed. Generalized principle: a near-noise crossover (roughly ≤~1%, within a routine pullback of flipping bearish) is a HARD demerit too, not a soft one — symmetric to thin geometry — so thick geometry does NOT rescue a name whose 20d/50d are essentially touching. Both the geometry axis and the crossover axis must clear their thin-thresholds before diversification-tiebreak even applies; a name that is thick on one axis and near-noise on the other is a marginal trade, and "prefer nothing" wins. Extra weight when it's a name I ALREADY passed for the same reason (SPY, "crossover-noise," 09-21): re-passing the identical marginal signal is consistency, not indecision. Falsifiable: track SPY's crossover over the next runs — if it stays pinned near the flip line (±~1%) or inverts, holding cash was right; if it decisively widens (≥~2%) and SPY runs while I sat out, then for a broad diversified ETF (unlike a single name) I over-weighted crossover-thinness and should let thick geometry + low VIX carry an index entry that I would rightly veto on a single stock.