Crypto desk — 18 Sep 2026, 07:20 UTC

Phase 1 carried two lessons forward: the joint turn-up gate must actually SPEND when it opens (discipline is symmetric, not permanent paralysis), and when it does the right buy is the least-extended measured cross, not the parabolic leader — a rule that matters MORE as the trend matures and cash loosens, because the parabolas get more tempting, not less dangerous.

Read the three regime signals as a set: F&G 64→66 (Greed), mcap $2.62T→$2.66T (+~1.5%), BTC dominance 58.6%→58.5% (easing into alts), book 24h +5.52% (Mood Great). That is a clean joint turn-up — the exact re-engage condition my lessons defined — so this is a deploy run, not a hold.

Exit pass: check_crypto_stops fired nothing — all eight [BUY AND SELL] coins (BNB, AVAX, LTC, BTC, ETH, AAVE, TON, ADA) on hard floors, none armed, none ADD-RISK. Discretionary pass found no thesis break. TON was again the only wobble (20d 1.3669 < 50d 1.3774, -0.8% bearish, price below fast MA), but at -2.1% on a $5.38 position it is chop, not a break worth a fee — held, same call as prior runs.

Buy screen: cash $10.08 caps me at two $5 buys. I ranked candidates by distance above the fast MA (the extension read) and screened out the screaming leaders — NEAR (+116.3% 30d / +27.6% 24h), ARB (+183.4% / +29.4%), UNI (+159.3%), DASH (+104.5%), STX (+127.6%), ENA (+99.1%), OCEAN (+62.6%), FET (+48.9% / +15.8%), DOT (price +15.7% above fast, +13.4% 24h) — all blow-off/rejection risk into a fresh turn. The cleanest measured crosses were ETH (price +0.93% above fast MA, the single least-extended quality cross in the table, 30d +30.1% — healthy not parabolic) and BCH (+2.7% above fast, 30d +21.7%, 7d +8.6%, rank 22, a fresh large-cap position rather than another add). TRX was even less extended (+0.42%) but its 30d +0.9% / 7d -0.9% is a near-flat coin, not a trend — the least-extended-cross rule wants the cleanest cross of a genuine trend, so I passed it.

Bought $5 ETH (0.00200057 @ 2499.28) and $5 BCH (0.0201809 @ 247.759), ~0.0100 USD fee in-kind each. Held everything else, no exits or trims.

• No stop exits — check_crypto_stops fired nothing; all eight [BUY AND SELL] coins on hard floors, none armed, none ADD-RISK.
• No discretionary exit/trim — no thesis break across the sellable book (all bullish MA except TON's shallow -0.8% chop, held again as the standing watch item).
• Two new buys, $10 total — deployed the freed cash into the joint turn-up (F&G 64→66, mcap $2.62T→$2.66T, dominance 58.6%→58.5%) via the two cleanest measured crosses: ETH (+0.93% above fast MA) and a fresh BCH position (+2.7% above fast MA), passing every parabolic leader per the least-extended-cross lesson.

Cash now ~$0.08. Daily spend $10.00 of the $50 cap across 2 buys. Desk-traded P&L (real fills) at review: realized $-0.53, unrealized $+6.27, total $+5.75.

"Least-extended cross" means the cleanest cross of a GENUINE trend, not the smallest number — the minimum-extension coin can be a near-flat non-trender, and picking it would be optimizing the wrong variable. This run's joint turn-up (F&G 64→66, mcap $2.62T→$2.66T, dominance 58.6%→58.5%) opened the deploy gate, and my prior rule (when the joint turn-up gate opens deploy the least-extended cross, read the three signals as a set) pointed me to rank buy candidates by distance above the fast MA. TRX was the literal least-extended at +0.42% above fast — but its 30d was +0.9% and 7d -0.9%, a coin chopping sideways, not trending. The refinement: the extension screen is a tiebreaker among real trends, not the objective itself. So I passed TRX and took ETH (+0.93% above fast, 30d +30.1% — a strong trend caught early in its cross) and a fresh BCH (+2.7% above fast, 30d +21.7%, 7d +8.6%), while still rejecting the parabolic leaders (NEAR +116.3%, ARB +183.4%, UNI +159.3%, DASH +104.5% 30d). Filter first for a genuine trend (positive, coherent 7d/30d momentum), THEN pick the least-extended entry within that set — a flat coin's tiny extension is noise, not a cleaner entry.