Crypto desk — 23 Oct 2026, 19:16 UTC

Phase 1 carried two live lessons: (10-22) when the tape is broadly red a lone green/above-fast name is a counter-trend outlier to distrust, hardest on memes; and (10-23) the flat-base filter is a green light only when a FRESH non-meme clears the MA gate AND its pop sits on a genuinely flat multi-week base.

check_crypto_stops fired nothing — APT (-8.9%, peak +5.5%), ARB (-9.4%, peak +4.2%), DOT (-5.0%, peak +0.2%), NEAR (-6.9%, peak +3.6%) all rest on their hard floors, none ADD-RISK-armed. Discretionary review of holdings: the only one with a wobble is DOT, bought yesterday at 1.1846, now 1.1232 — the flat-base reclaim I paid for has slipped back below its fast MA (1.1651). But it's a shallow -5%, the 20d>50d cross is still bullish (+3.1%), and the backdrop is neutral not risk-off (F&G 48, BTC dominance flat at 60.0%). That's a hold-on-the-hard-floor situation, not a thesis-broken dump; churning out at -5% to pay another fee isn't warranted. APT/ARB/NEAR likewise down only on broad alt softness, MAs still bullish. No discretionary exit.

Buy side: three Candidate = buy names — PEPE, MEW, BCH. PEPE is a Meme up +7.5% 24h / +20.2% 7d / +8.5% 30d with a "rises 11% daily, pump or correction?" headline — a textbook extended blow-off, exactly what my momentum lesson says to fade. MEW (+3.6% 24h / +17.6% 7d) is the same story, a stretched counter-trend pop -96.2% below ATH. Both skip. BCH is the only non-meme candidate: fast 284.8634 > slow 278.6926 (+2.2%), price 285.135 — but that's only ~0.1% above the fast MA (a marginal, not decisive, reclaim) and its 30d is -18.2%, a real downtrend with a recent bounce, NOT the flat base that made DOT a buy yesterday (+2.3% 30d, decisively above fast). BCH fails both quality tests my lesson set for choosing DOT over ATOM. The disciplined read: a barely-there reclaim on a 30d downtrend is a weak signal to let prove itself, not chase.

Held all six positions. No exits, no buys. Stop pass triggered nothing; no holding warrants a discretionary exit (DOT's faded reclaim is a shallow -5% on a still-bullish MA and a neutral tape — hold on the hard floor, don't churn). Of the three buy candidates, PEPE and MEW are extended memes I fade on principle, and BCH is a marginal fast-MA reclaim (+0.1% above) sitting on a -18.2% 30d downtrend — it fails the flat-base quality bar DOT cleared yesterday, so I pass rather than pay a fee to chase a weak signal. Within caps regardless: $5.00 of $50 daily spent, $45.00 left, 0 of 3 new buys used. Cash $10.28, book value $28.72 at review; desk-traded P&L realized (30d) +$0.00, unrealized -$1.58, total -$1.58.

The flat-base buy signal has TWO quality gates, not one: the reclaim must be DECISIVE (price meaningfully above the fast MA) AND the multi-week base must be genuinely flat (near-zero/positive 30d), not a bounce off a 30d downtrend. A Candidate = buy that clears the MA cross but fails either gate is a weak signal to let prove itself, not chase. This run BCH flagged buy (fast 284.8634 > slow 278.6926, +2.2%; +4.1% 24h / +3.9% 7d) — but price 285.135 was only ~0.1% above the fast MA (a marginal reclaim, not the decisive one DOT had), and its 30d was -18.2% (a downtrend with a recent bounce, NOT a flat base). Contrast the DOT buy I took yesterday: price decisively above fast and 30d +2.3% — both gates passed. Same MA-cross flag, opposite quality. I passed on BCH and the two other candidates (PEPE +20.2% 7d / MEW +17.6% 7d, extended memes to fade). Generalize: after the MA gate, grade the reclaim's decisiveness and the base's flatness separately; a barely-above-fast pop on a monthly downtrend is the laggard-bounce you should distrust, even in a neutral tape (F&G 48, dominance 60.0%). Held all six; no stop fired; unrealized -$1.58.