Crypto desk — 16 Sep 2026, 07:20 UTC

Phase 1 lessons first. Both recent lessons converge on the same discipline: the fresh-joint-turn-up gate decides whether to deploy at all, and flat/rising BTC dominance means there is no risk-on-for-alts tell — so an above-fast-MA screen is necessary but not sufficient. A persistent down-drift is not a dip-buy signal no matter how many coins screen above their fast MA.

Read the regime through those lessons. F&G ran 70→67→64 over the last three runs and now sits at 63 — still ticking down, no fresh turn-up. Total mcap $2.59T→$2.58T (down again). BTC dominance flat at 59.0% (not easing). Book 24h -4.34%, mood "Rough". This is a fourth consecutive down-drift reading, the exact mirror of the re-engage trigger — the gate is shut. And with dominance flat, even if the gate were open the instrument would be the major, not an alt.

The four buy candidates — NEAR (price 2.35 vs fast MA 2.1865), ARB (0.153256 vs 0.13589), UNI (6.29 vs 6.0697), OCEAN (0.155069 vs 0.147865) — all screen above their fast MAs, which in isolation reads "deployable." All four are alts, so with dominance flat they fail the instrument test regardless. More decisive: cash is at the $5.06 floor and the per-buy cap is $5, so any single buy drains the entire powder to ~$0.06 into a declining tape. That is exactly the trap the 09-15 19:20 lesson warns against. Prefer doing nothing over a marginal trade. News (BTC ETF $450M outflow, BTC at a September low on global bond stress) is attention only and reinforces the risk-off backdrop — not a reason to act.

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 (BTC, ETH, BNB, AAVE, TON, ADA) on hard floors, none armed, none ADD-RISK.
• No discretionary exit/trim — no thesis break across BTC/ETH/BNB/AAVE/ADA (all bullish MA). TON remains the standing watch item: bearish cross (20d 1.3684 < 50d 1.3802, -0.9%) and flagged sell in research, but it is a shallow -5.5% chop on a $5.21 position — a break not worth a fee, held again.
• No new buys — regime is a fourth consecutive down-drift (F&G 64→63, mcap $2.59T→$2.58T, dominance flat 59.0%), so the re-engage gate is shut; dominance flat also kills the alt-rotation tell, and the only candidates (NEAR, ARB, UNI, OCEAN) are alts. Cash at the $5.06 floor would be fully drained by a single $5 buy into a cooling tape — I preserve the powder.

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 $-12.22, total $-12.75.

When cash is pinned at the per-buy floor, a "buy" is not a tranche — it is all-in, so the bar rises rather than falls, and repeated candidate flags across a down-drift are noise, not accumulating evidence. This run cash was $5.06 against a $5 per-buy cap, so any single buy would have drained the powder to ~$0.06. The regime was a fourth consecutive down-drift (F&G 70→67→64→63, mcap $2.59T→$2.58T, dominance flat 59.0%), and the SAME alt candidates that flagged last run (ARB, UNI, OCEAN, NEAR) all screened above their fast MA again — but a candidate set that persists unchanged across a shut gate is the MA-screen repeating itself, not a signal strengthening toward action. Generalized principle, extending the regime gate dominates the per-coin above fast MA screen: two conditions each independently veto a buy here — (1) a still-shut turn-up gate on a persistent down-drift, and (2) cash at the per-buy floor, where "deploy" means committing the last of the powder in one shot into a declining tape. When both hold, the disciplined move is not to wait for the candidate list to "finally" earn a buy — it is to recognize the repeat as confirmation to keep holding until the joint readings turn back up together.