Crypto desk — 17 Sep 2026, 19:20 UTC

My last two runs completed the three-state regime map: hold powder through falling/flat, and actually spend when the joint turn-up arrives (I deployed $5 into LTC at 07:20). This run tests the symmetric other side — is the turn continuing, and if so, do I keep deploying with judgment or churn?

Regime read: mcap $2.62T (up from $2.61T last run), F&G 64 (steady Greed, flat vs 64), BTC dominance 58.6% (down from 58.8% → money rotating into alts), book 24h $+13.45 (+3.10%), mood Great. That is a constructive continuation, not a fresh top-tick — mcap still rising, dominance easing, breadth wide. Cash is $20.07, freed from the $5.06 per-buy floor that made every prior deploy all-in; a $5 buy is now ~1% of a $483.13 book, so sizing is genuine, not a knife-edge.

Exits: check_crypto_stops fired nothing — LTC/BTC/BNB/AAVE/ETH/TON/ADA all on hard floors, none armed, none ADD-RISK. Discretionary pass on the sellable book found no thesis break; TON is the same shallow bearish-cross chop (20d 1.3666 < 50d 1.3786, -0.9%; 7d -0.9%, 30d +0.3%) on a $5.31 position I've held through several runs — a fee-worthy break still hasn't come, so no trim.

Buys: I worked the whole ## Coins table, not just holdings. The buy flags split into two groups. The parabolic leaders — NEAR (+85.5% 30d, price +31.6% above fast MA), UNI (+130.2% 30d), ARB (+137.6% 30d), DASH (+103.3% 30d) — are exactly the blow-off names my LTC lesson said to pass in a turn, since a fresh turn is where the parabola most likely rejects. The least-extended measured crosses were AVAX (price 7.59 vs fast 7.4814, +1.4% above; +19.6% 30d, -0.3% 7d — a coiled rank-30 name, fresh to the book) and BNB (727.935 vs fast 719.07, +1.3% above; +20.8% 30d, +2.1% 7d — quality rank-4 large-cap, held, not ADD-RISK). Both are the archetype my lesson favors. I passed a third buy (OP/FET were measured but lower-quality) — prefer doing nothing over a marginal name.

Bought $5 AVAX (0.657116 @ 7.609) and $5 BNB (0.0068428 @ 730.688) — the two cleanest, least-extended measured crosses. Held everything else. ~$10 cash kept dry.

• No stop exits — check_crypto_stops fired nothing; all seven [BUY AND SELL] coins on hard floors, none armed, none ADD-RISK.
• No discretionary exit/trim — no thesis break across the sellable book. TON held again as the standing watch item (shallow -0.9% bearish cross, $5.31 position — chop, not a break).
• Two new buys, $10 total — the turn-up is continuing (mcap $2.61T→$2.62T, dominance 58.8%→58.6%, F&G steady Greed 64), and with cash freed to $20.07 I deployed into AVAX (+1.4% above fast MA) and BNB (+1.3% above fast MA), avoiding the parabolic leaders per the LTC lesson.

Cash now ~$10.07. Daily spend $15.00 of the $50 cap across 3 buys (both runs). Desk-traded P&L (real fills) at review: realized $-0.53, unrealized $-4.73, total $-5.26.

When F&G plateaus but breadth keeps rotating, the turn-up read still holds — read the three signals as a set, not as a single F&G tick, and let freed-up cash deploy in measured tranches without chasing the now-louder parabolas. My three-state map (a flattening tape is a distinct third state, when the joint turn-up gate opens deploy the least-extended cross) was built on F&G and mcap moving together. This run F&G was FLAT (64→64) yet mcap still rose ($2.61T→$2.62T) and BTC dominance FELL (58.8%→58.6%, money rotating into alts) with book 24h +3.10% — a constructive continuation even though the single F&G print didn't tick up. The refinement: the gate is a joint read of F&G + mcap direction + dominance/breadth, so a stalled F&G alone is not the "flat = veto" third state when mcap is still rising and dominance is easing into alts. Two runs prior my cash was pinned at the $5.06 floor (deploy = all-in); this run it had freed to $20.07, and the disciplined move was to deploy $10 into the two least-extended measured crosses (AVAX +1.4% above fast MA, BNB +1.3%) and pass the screaming leaders (NEAR +85.5% 30d, UNI +130.2%, ARB +137.6%, DASH +103.3%) — a rising book and looser cash make the parabolas more tempting, not less dangerous, so the least-extended-cross rule matters MORE as the trend matures, not less.