IDEAL RESTRUCTURE, THREE-ACCOUNT PLAN, P&L AT EXPIRY MODEL

Gross-long market exposure (shares + long-call underlying) bucketed by theme, per account and for the whole book. Main is the diversified income core, Joint is index beta, Neville is the crypto convexity sleeve; Joint is $0.35M of USD cash (its own solid grey pie). The grey Cash slice in every pie is that account's idle reserve. The TOTAL pie is the honest picture of where the concentration sits.

The honest denominator. The pies above are market exposure (leveraged notional); these are capital — every pie sums to the account's real equity, so cash shows its true weight. Deployed capital = invested capital (net debit) bucketed by theme.

RetiredInc is CLOSED post-liquidation: its index book and its cash both moved into Joint, joining the prior $0.35M balance — Joint is now ONE $1.35M account (index fortresses + ~$0.81M cash; ML $0.94M = 0.70x cash, a DEEPER buffer than the original 0.94x sizing). Smart-capped: low-vol income/ballast names (JPM, XOM, LMT, LLY) plus GOOG, AMZN, SNDK are fully written for income; the high-IV/high-beta names AND GLD stay half-uncapped for convexity (GLD moved to half-capped after its Aug V-bounce ratcheted past the ATM call); SPY is a LEAPS. Validated against the Margin & Liquidation Field Guide: Main and Joint both 0.94x ML/cash; crossover "none" for BOTH accounts. Neville (the $0.65M crypto sleeve) is shown separately below — its 2b Fortress carries no income and no margin, but its $513k idle cash reserve (79% of the sleeve, backing ML < cash) DOES earn the idle-cash rate, so that interest IS folded into the income card above.

Unrealized P&L of each book if every underlying is haircut the same % (instantaneous shock, crash-vol adjusted: IV rises as price falls, which lifts the long hedge puts and adds to short-put liability). Structure legs only (LC + SP + HP); income CCs expire worthless into a selloff and are not counted, so the real cushion is slightly better. Neither account is force-liquidated anywhere in this range (crossover "none" past −95%), so these are paper drawdowns the book rides, not liquidation events.

The mirror of the drawdown chart: mark-to-market GAIN on the same LC/SP/HP structures as every underlying rallies. Cap-aware — the written (income) contracts are capped by their covered calls near one cycle; the uncapped half and SPY run free. The pale line is the uncapped ceiling (stop writing / roll caps up to reach it).

v28-recommended archetype per ticker. NVDA dropped (2026-08-24) — it was 47% duplicated via QQQ and only 25%/yr inc/ML; its max-loss upsized QQQ (11→15), which already holds NVDA ~9%. GOOG and AMZN are now fully written (not half): they are low-IV mega-caps with little upside convexity to protect, so writing every contract converts that dead upside into income (~2x each: GOOG $2.1k→$5.0k, AMZN $2.5k→$5.8k/mo). Low-vol income/ballast names (JPM, XOM, LMT, LLY, GOOG, AMZN) are fully written; the genuinely high-IV/high-beta names (MU, AMD, semis, crypto) keep half uncapped for convexity.

2026-08-24 merge: RetiredInc closed; this index book and $1.35M of cash now live together in Joint. ML $944,947 = 0.70x cash; battery re-run at this sizing: crossover NONE past −95% under every attack, worst path NLV $622,024, cash−ML floor $405,053. "Book B" (cap-policy rebalance, 2026-08-23): QQQ is kept, but half-capped. The first pass dropped QQQ as redundant with SPY; that was right about exposure and wrong about function — QQQ was the account's income engine and SPY structurally cannot replace it (SPY's ~16% IV is exactly why it is held as an uncapped LEAPS rather than written). The real defect was never holding QQQ, it was modelling it as written at-the-money, which the live book does not do. So contract counts are restored, then QQQ upsized to absorb dropped-NVDA's ML (SPY 9 / QQQ 15 / GLD 29; IGV was subsequently dropped, see the XBI + reserve note below). Cap policy: QQQ and GLD are half-capped. Each can run (gold V-bounced in Aug and ratcheted past its call; software and Nasdaq can do the same), so half of each is left uncapped for convexity while half is written for income. Income lands at ~$14.8k/mo versus the $25.8k the fully-written ATM model claimed — the difference was never real. Note the conservatism: half-capping costs more than half the income, because the hedge bill is charged on ALL contracts while premium is credited only on the written half. 2026-08-24 — XBI healthcare-AI sleeve + $350k withdrawable reserve: a 26-contract LeapBuilder synthetic long (Bucket A: LC 145 / SP 178 / HP 160, Dec-2027, ~4.8x, breakeven ~$179 vs spot $166), a convexity bet on biotech being positively disrupted by AI. To keep ~$350k always withdrawable-and-safe, Joint's ML is held at ~$1.0M so the cash−ML floor equals the reserve: even an all-to-zero leaves the reserve intact and pullable at any time. That budget was made by dropping IGV entirely (redundant software vs QQQ + Main's AI cluster, thinnest option chain in the sleeve) and sizing XBI to 26. CC overlay (2026-08-24, live TWS): 15 of the 26 contracts write a near-money 0.20-delta call ($175, ~$1.15 credit, ~monthly, rolled) — biotech's high IV makes this rich, ~$4,774/mo gross / ~$3,246/mo net (keep ~68% at 0.20 delta, an ESTIMATE: no XBI trade history to calibrate). The other 11 contracts stay fully uncapped forever, and all 26 keep their upside up to $175. This is the deliberate trade: 15 contracts capped at ~+5% harvest the fat premium (get paid to wait), 11 run free for the disruption tail. Two caveats: (1) the $175 strike is ~$4 UNDER the LeapBuilder CC-SS of $178.79, so the tool's model nets slightly negative on assignment — covered in practice by the deep-ITM $145 LC on a short-dated diagonal, but writing $179+ stays strictly above the safe floor at ~the same monthly premium; (2) writing near-money means diligent monthly rolling — a fast breakout caps the 15 until you roll up. Net Joint: SPY / QQQ / GLD income core + XBI convexity-plus-income $3,246/mo, ML ~0.74x of the $1.35M, crossover NONE past −95%; the short calls add ~$30k of maintenance margin (not ML), easily inside Joint's headroom.

Crypto held as the 2b Fortress (see the Crypto tab for the full existing-book, moonshot and fortress analysis): leap_builder-optimized held-hedge calls (long call + short-put-spread per name) reproducing the book's crypto upside (~$26.0M at the end-2027 targets) as a DEFINED-RISK structure. Net cash ~$137k; worst-case structure max loss $632,559, held BELOW the $650k cash so ML < cash — it cannot be force-liquidated and does not touch Main/Joint's liquidation math. This is the isolated crypto tail bucket; it substitutes for layering more crypto into Main.

Crypto is held as the 2b Fortress (leap_builder-optimized held-hedge calls: long call + short-put-spread per name). Full legs, strikes, heatmaps and the moonshot alternative are on the Crypto tab. Neville is funded to $650k cash specifically so the structure max loss ($632,559) sits below cash — ML < cash, so it cannot be force-liquidated.

In the black by BMNR 30 / MARA 20 (+26%); breakeven ~BTC +14%; scales ~linearly with option premium. The four names are all Crypto & Fintech, which is why the TOTAL pie's amber slice grows once Neville is added — the deliberate, isolated, defined-risk crypto tail. MARA is now EXECUTED (2026-08-20): 10,000 shares + 400× Jan-28 13C, currently held in Main (to be moved into this sleeve at the redeploy; it consumes ~$90k of Main margin until then). BMNR/COIN/IBIT remain proposed.

The honest alternative: buy the same names outright with the same $3,000,000 and write covered calls on them. Priced entirely off realized covered-call yields from the trade log, not the planning model, on BOTH sides. This is what the fortress structure is actually buying you, and what it costs. (Joint's $350k withdrawable reserve is excluded from both sides — it sits aside as cash whether the book is fortresses or stock, so comparing on the deployed $3.0M is the fair basis.)