Crypto desk — 15 Sep 2026, 19:20 UTC

Phase 1 read first. Cash is $5.06 — barely one $5 buy — book value $446.43, 24h -6.01%, mood "Rough," desk-traded P&L realized $-0.53 / unrealized $-10.21 / total $-10.74. No open orders. Daily spend $20.00 of $50 across 1 buy, $30.00 left.

The regime is the deciding factor. Track the last three runs: 09-14 19:20 F&G 70 / mcap $2.69T (deployed $5 BTC on a fresh joint turn-up), 09-15 07:20 F&G 67 / mcap $2.63T (held — softening), and now F&G 64 / mcap $2.59T, dominance 58.9% flat. That is a third consecutive down-tick on both readings — a fresh joint turn-down, the mirror image of the re-engage trigger. My lesson the re-engage tell is a fresh turn-up not a steady firm level gates deployment on the readings incrementing UP together; here they are incrementing down. And dominance at 58.9% is flat, not easing, so per falling dominance is the risk-on-for-alts tell there is no alt-rotation signal either. Both gates fail.

Research candidates flagged buy: BNB, NEAR, ARB, UNI, OCEAN — all above their fast MA. But an above-fast-MA screen is necessary, not sufficient; the regime gate has to be met too, and it isn't. Every coin in the book is red 24h and the whole tape is -6.01%. Deploying my last $5.06 into a declining, risk-off tape with no fresh turn-up is precisely the marginal buy CLAUDE.md tells me to skip — and it would drain the powder to ~$0, exactly the drift my re-engage lesson warns against.

Exit passes: check_crypto_stops fired nothing — BTC, ETH, BNB, AAVE, TON, ADA all on hard floors (BTC peak +4.2%, ETH +6.2%, BNB +12.1%, AAVE +6.7%, TON +5.1%, ADA +3.9%), none armed, none ADD-RISK. Discretionary pass on the [BUY AND SELL] names: no thesis break. BNB is healthy (now +3.6%). ETH/BTC/AAVE/ADA are shallow red on bullish MAs, above their floors — a normal pullback, not a break. TON stays the standing watch item: research flags it sell on a bearish cross, but it's only -0.6% (20d 1.3736 < 50d 1.3824) on a $5.24 position down -4.7% — shallow chop that isn't worth a fee to churn, consistent with how I've held it the last two runs.

Placed no orders. Held the entire book. Keeping $5.06 dry.

• No stop exits — check_crypto_stops fired nothing; all six [BUY AND SELL] coins on hard floors, none armed, none ADD-RISK.
• No discretionary exit/trim — no thesis break across BTC/ETH/BNB/AAVE/ADA. TON held as the standing watch item (marginal -0.6% bearish cross on a $5.24 position, -4.7% — chop, not a break).
• No new buys — the regime turned down again, not up: F&G 67→64, mcap $2.63T→$2.59T, dominance flat at 58.9%. A fresh joint turn-down with dominance not easing, so neither the re-engage gate nor the alt-rotation gate is met. Candidates above their fast MA (BNB, NEAR, ARB, UNI, OCEAN) don't change that — I preserve the last $5.06 rather than force a marginal buy into a rough, cooling tape.

Cash stays $5.06. Daily spend today $0.00 of the $50 cap across 0 buys. Desk-traded P&L (real fills) at review: realized $-0.53, unrealized $-10.21, total $-10.74.

The regime gate dominates the per-coin "above fast MA" screen: when the joint readings are turning DOWN, an above-fast-MA candidate is not a buy — the screen is necessary, not sufficient, and a down-drifting tape vetoes it. This run five candidates sat above their fast MA (BNB, NEAR, ARB, UNI, OCEAN), which in isolation reads "constructive, deployable." But the regime had ticked down a THIRD consecutive run — F&G 70→67→64, mcap $2.69T→$2.63T→$2.59T, dominance flat 58.9%, book 24h -6.01% "Rough" — a fresh joint turn-down, the exact mirror of the re-engage trigger from the re-engage tell is a fresh turn-up not a steady firm level. With cash at the $5.06 floor, buying any of those names would have drained the powder to ~$0 into a declining tape. Generalized principle: the fresh-joint-turn-up gate decides whether to deploy at all; the coin-level MA screen only chooses among candidates once that gate is open. A persistent down-drift is not a dip-buy signal, no matter how many coins screen above their fast MA — hold the powder and wait for the readings to turn back up together (pairs with falling dominance is the risk-on-for-alts tell for the instrument choice once they do).