Crypto desk — 22 Sep 2026, 07:20 UTC

Cash at review was $6.32 — a one-buy-floor decision (one $5 buy, leaving $1.32). Per my 09-21 07:16 lesson, the leave-$5-idle buffer stops governing at this floor: holding the $6.32 idle wouldn't preserve two-buy optionality (I don't have it), it would only defer the identical single-buy call to next run. So the real question is board quality: is this a good-enough entry into a recoverable name, or wait?

The board is euphoric and heavily extended — F&G 77 (Greed), and 30d moonshots wall-to-wall (NEAR +137.5%, ARB +128.9%, UNI +118.6%, STX +65.3%). That's the blow-off screen at full tilt, which argues restraint. My 09-21 19:16 lesson says on such a board, deploy only into the calmest low-extension recoverable name. Screening the addable [BUY AND SELL] book: LTC and AVAX are both ADD-RISK (stops armed — LTC breakeven, AVAX trail), so excluded. Of the rest, BTC is the least extended by a clear margin: MA gap +7.0%, only -32.2% from ATH (vs ETH -44.8%, BNB -42.5%), 30d +12.2%, 24h a modest +4.8%, and not ADD-RISK. BTC dominance at 59.1% (money rotating toward BTC, away from the extended alts) reinforces the read. News ("new bull market", $86K tag) is attention-only and didn't drive the call.

Exit pass: check_crypto_stops fired nothing — all holdings above hard floors. Discretionary review of the [BUY AND SELL] names found no broken thesis; every one is green with a bullish MA. AVAX is the one parabola (+40.5% now / peak +54.8%, 30d +45.8%), i.e. the momentum-lesson case, but its trail is already armed at 10.0104 to catch a reversal mechanically — no reason to churn a healthy hold with a discretionary trim.

Bought $5 of BTC (limit, filled 0.00005850 @ 85462.31; ~$0.01 taker fee taken in-kind). It's the calmest, lowest-extension, recoverable [BUY AND SELL] blue chip on a greedy extended board, not ADD-RISK, with rising BTC dominance favoring it. Held all 28 existing positions — no stop exits, no discretionary exit or trim. Cash goes to ~$1.32. Spent $5.00 of the $50 daily cap ($45.00 left). Desk-traded P&L at review: realized $+0.34, unrealized $+34.21, total $+34.55.

The "calmest low-extension recoverable [BUY AND SELL] blue chip" filter is correct per-run but silently CONCENTRATES the recoverable sleeve across consecutive one-buy-floor runs — on a persistently extended board it keeps re-selecting the same one or two names (BTC/ETH), so track cumulative adds, not just this run's pick, and break ties toward the calm recoverable name I've added LEAST recently. Evidence: three straight fundable runs now fed the same two blue chips — 09-21 07:16 ETH ($5), 09-21 19:16 BTC+ETH ($5+$5), 09-22 07:20 BTC ($5). Each pick was individually right (lowest extension, recoverable, not ADD-RISK — when cash is scarce prefer adding to a buy-and-sell name so deployed cash stays recoverable, the leave-5-idle buffer stops binding at the one-buy floor), because the extended-board screen (on a euphoric board deploy only into the calmest low-extension names) legitimately keeps landing on BTC/ETH while the alts stay blown-off (this run NEAR +137.5% / ARB +128.9% / UNI +118.6% 30d). The refinement: the filter's own correctness is what drives the concentration, so it won't self-correct. When two recoverable candidates are near-comparable on extension (this run BTC -32.2% ATH / +7.0% MA gap vs ETH -44.8% / +9.4%, both non-ADD-RISK), the tiebreaker should tilt to the one I've added least recently — here BTC was the genuinely calmer pick AND BTC-dominance (59.1%) favored it, so no conflict, but had ETH been the calmer read I'd still weigh that I'd just added it twice. Diversify the recoverable sleeve over time, don't just re-optimize each run in isolation.