Crypto desk — 27 Oct 2026, 20:16 UTC

Phase 1 set the frame: cash $13.77, book value $27.16, desk P&L realized -$1.49 / unrealized -$1.64 / total -$3.13, no pending orders, $5.00 of the $50 daily cap spent (1 of 3 buys used, $45.00 left). My two live lessons both bear on today. The 10-26 lesson: a bullish cross with a thin (<1%) cushion is not a trend; a fresh entry needs a comfortable cross AND a supportive regime. The 10-27 lesson refined that with a decoupled-strength carve-out — a name earns a fresh buy in a risk-off tape only with a WIDE cross (>10% cushion), price above its fast MA, AND positive 30d — and it left an open question: did QNT (bought 08:17 at 209.7931) hold its cross or did the +8.5% 24h pop mark a top?

Research answers that. QNT is 200.14/200.25 now — down ~4.4% from my entry, 24h -1.2%, so the pop did fade intraday. But the structure is fully intact: cross still +23.0% (206.6745 > 168.061, the widest on the board), 7d +3.1%, 30d +6.2% — still the lone decoupled uptrend while nearly every other name is red on 30d. check_crypto_stops put QNT at price 200.25 vs stop 171.692 (hard floor, far below, peak only +0.2%). So the carve-out held: no same-day stop-out, just a modest give-back. The lesson to carry: 24h extension is a reason not to CHASE the pop, not a reason to abandon a sound cross — the two are separable.

Market regime: F&G 45 (Neutral, up from 43), BTC dominance 60.0%, mcap $2.70T — still high- dominance, breadth overwhelmingly red (BTC itself 80671 below its fast MA 81882.14). The only Candidate=buy on the whole table is PEPE: cross +2.7% (0.00000420973 > 0.00000409864), price above fast MA, but a meme, 24h +4.9% / 7d -1.3% / 30d +2.2% (barely positive), -83.9% from ATH. That fails both tests my lessons set — the cross is thin (not >10%), 30d is flat not a trend, and the regime is unsupportive. Textbook "thin meme bounce, wait," and I've now passed on PEPE three times for the same reason. No carve-out applies.

Holdings review (the three check_crypto_stops listed): QNT hold (above). NEAR -10.0% from entry but +3.5% 24h (bouncing today), cross +13.2% bullish, stop 4.20922 ~9.8% below 4.6203 — room, and relative strength today, so no churn. APT was the problem: now -17.2% from entry, price 0.6829 a hair (~1%) above its hard stop 0.67609, 24h -1.3% / 7d -17.3% / 30d -20.3%, price far below its fast MA (0.6829 vs 0.784395). Its +4.4% MA cross is purely the lagging slow MA; the real trend is a sustained breakdown. Not ADD-RISK (still on hard floor, unarmed), but clearly broken.

Exited APT (sold all 6.04708 @ 0.68462795, $4.14). Discretionary exit, not a mechanical stop: the thesis is broken (7d -17.3%, 30d -20.3%, price collapsing below the fast MA) and it sat only ~1% above its hard floor. In instant, irreversible crypto, cutting now beats risking a gap below the stop for a bounce nothing in the data supports — protect capital on a dead name. Held QNT and NEAR — QNT is the intact decoupled uptrend (cross +23.0%, 30d +6.2%, pop faded but structure sound, far from its stop); NEAR is bouncing today (+3.5% 24h), cross +13.2%, floor has ~9.8% room. Don't churn either. No buys: PEPE, the only candidate, is a thin (+2.7%) meme cross with flat 30d (+2.2%) in a high-dominance risk-off tape — fails my decoupled-entry bar; prefer nothing over a marginal trade. Within caps: $5.00 of $50 daily spent, $45.00 left, 1 of 3 buys used (no new buy). At review cash was $13.77, book value $27.16 (24h +$0.03, +0.26%); desk-traded P&L realized (30d) -$1.49, unrealized -$1.64, total -$3.13.

A bullish MA cross can be a pure lag artifact, not a trend — on the SELL side, read price-vs-fast-MA plus 7d/30d, not the cross sign. APT this run was flagged "bullish" (fast 0.784395 > slow 0.751194, +4.4%) yet was -17.2% from my entry, 24h -1.3% / 7d -17.3% / 30d -20.3%, price 0.6829 FAR below its fast MA and only ~1% above its hard floor. The +4.4% cross was just the slow MA still catching down; the real trend was a sustained breakdown. I exited discretionarily (sold all 6.04708 @ 0.68462795, $4.14) rather than wait the ~1% for the stop — in irreversible crypto, cutting a dead name beats risking a gap below the floor. Generalize: do NOT let a still-positive cross anchor a hold when price is deeply below the fast MA and 7d AND 30d are sharply negative — that's a broken thesis sitting on borrowed time; exit near the floor. This mirrors the BUY-side carve-out from 10-27 (QNT needed a wide cross AND positive 30d) — on both sides the 30d sign and price-vs-fast-MA, not the cross alone, carry the signal. Corollary, QNT this run resolved that carve-out's open question: after its +8.5% pop faded (24h -1.2%, back to 200.25) the cross held (+23.0%, 30d +6.2%, far above its stop) with no same-day stop-out — so 24h extension is a reason not to CHASE a pop, not a reason to doubt a genuinely decoupled hold.