Tax Architecture & ROC Strategy

Part 3 chose the convexity backbone. Part 4 asks the next question: once you own convexity, how much of it do you get to keep? Sell one large, long-held winner from a taxable account and taxes can take 15 to 35 percent or more of the gain, depending on income, the net investment income tax (NIIT) and state tax; realize gains every year along the way and the bill is larger. In a taxable account the tax code is a silent partner: no say in what you buy, and a cut of every gain you take. Wrapper engineering routes each dollar to the wrappers that minimize lifetime tax friction and keep right-tail outcomes intact.

Conventional guidance says “maximize the deduction”: take pre-tax first, treat Roth as a side account for younger savers, and fill taxable once the limits are full, an order built to shrink this year’s tax bill. Wrapper engineering is built for the number you end with. Over decades, especially when gains are realized along the way, tax treatment can rival what asset selection adds, and sometimes exceed it. So the framework treats wrapper placement as a structural return multiplier and gives it the same rigor as asset selection and position sizing.

Jurisdiction scope. The examples use United States accounts: Roth accounts, the taxable brokerage account, and pre-tax plans such as a traditional 401(k) or IRA. The logic carries to other countries once you substitute the local tax-advantaged accounts: put the positions you expect to sell where selling costs least, and spend scarce tax-advantaged room on the holdings that gain most from it.

Among the thinkers the framework draws on (Part 2), Edelman makes the long-horizon case for why tax treatment matters, and none of them formalizes wrapper routing; the framework fills that gap deliberately.

Each of the three kinds of account has a job, and one rule assigns them: match what a position is expected to do (the shape of its payoff and how likely you are to sell it) to the account whose tax treatment rewards that behavior.

One placement looks backward at first: by default, the convexity backbone stays out of the tax-free account. Bitcoin, held in cold storage under a never-sell rule, belongs in taxable; the Roth section below gives the reasons, and Part 3 charts its path from accumulation to borrowing. Everything else follows a sequence for each new dollar, with no fixed split.

The sequence governs where each new dollar goes. Separately, the dashboard’s Framework Integrity Score (Part 6) measures the household’s dollar split against the FIS wrapper targets of 45 percent Roth, 35 percent taxable and 20 percent pre-tax, and charges a small Allocation penalty (at most about 6 of its 100 points) once the split drifts more than 10 points from them (as of September 2026; the formula is on the Math page). A book routed by this sequence can show that drift: the 45/35/20 split is a scoring parameter, not a Part 4 rule.

Roth capacity is scarce. The 2026 IRA limit is $7,500 per person ($8,600 at 50 or older), shared across traditional and Roth IRAs. It caps the Roth IRA, not Roth space as a whole: a Roth 401(k) takes up to $24,500 of your own deferrals in 2026 (more from age 50, with catch-up contributions), and where the plan allows after-tax contributions and converts them, a mega backdoor can add more, up to the $72,000 federal limit on everything added to the plan in a year (your deferrals and the employer’s contributions count toward it).

Compound interest has been called the eighth wonder of the world. Tax-free compound interest is the ninth. The framework’s cornerstone allocation principle follows: spend scarce Roth room on the highest-conviction Torque that may rotate, where tax-free compounding and tax-free trading do the most work. Consider the mathematics.

The gap between what each wrapper lets you keep is small on an ordinary gain and wide on a right-tail one, so placement matters most exactly where convexity is largest. The comparison below works the case of a gain realized once; Part 5 works the case of gains realized every year, and later Parts point back to these two.

A Roth keeps essentially all of a gain, since qualified withdrawals are not taxed. A taxable account that sells keeps roughly 75 to 85 percent of the gain after long-term capital gains tax and the NIIT, depending on income; the exhibit uses the top federal rate of 23.8 percent, the rate a large one-time sale meets. A pre-tax account keeps roughly 60 to 80 percent of the whole withdrawal, principal included, because every dollar comes out as ordinary income; the exhibit uses a blended 30 percent for withdrawals spread over retirement. On a position that grows tenfold, the Roth ends about 27 percent ahead of the taxable account that sells it, and on a single sale that lead never passes about 31 percent however large the gain. Hold the taxable position until death instead, and under current law a stepped-up basis can close the gap almost entirely. All of these figures are federal and illustrative; they depend on income, filing status, state tax and the rules in force when you sell or withdraw.

One caveat: the exhibit leaves out the deduction a pre-tax contribution earns going in. At the same tax rate going in and coming out, a Roth and a pre-tax account leave you the same after-tax money, whatever the return. Four things tip convex, right-tail positions toward Roth. Contribution limits are set in dollars, so a full Roth contribution shelters more after-tax money than a full pre-tax one. A large pre-tax balance comes out through required distributions, which can push it into higher brackets than the deduction saved. The original owner never takes required distributions from a Roth. And heirs generally inherit a Roth free of income tax under current law. In every case, the gap that widens as the win grows is the one between a Roth and a taxable account that sells.

Roth also eases three retirement risks: a withdrawal in a high-tax year, a guess about tax rates decades out, and a forced withdrawal in a downturn. Under current law, qualified withdrawals are tax-free whatever your income. Early Roth IRA withdrawals have rules of their own: contributions come out tax-free at any time, earnings are tax-free once you are 59 and a half and five tax years have passed since the year of your first Roth IRA contribution, and each conversion starts its own five-year clock for the 10 percent additional tax on converted money withdrawn before 59 and a half.

Why fill the Roth IRA before putting more Roth money into the workplace plan? Because a tax shelter is only as useful as what it can hold. Plan menus usually offer a short list of funds, while a Roth IRA at a brokerage can hold the individual positions and rotations the framework relies on. Position sizing, diversification and rotation mechanics are in Part 5.

A Roth may hold long calls, long puts and defined-risk spreads, never naked options, short volatility or any other undefined-risk strategy. Purchased options are tactical overlays for a high-conviction thesis, with the loss limited to the premium paid; they are neither core holdings nor income generators. Much of the time their expected value is negative, because time decay usually costs more than the moves repay, so they earn a place only when conviction justifies a defined-risk, asymmetric bet. Two representative uses: a long-dated call on an infrastructure theme early in its adoption, or a long put as portfolio insurance during systemic stress.

Taxable accounts hold what does not need Roth’s tax-free trading. Yield never qualifies a position as Ballast; Part 5’s tests do. Among holdings that pass them, the taxable account favors those whose distributions are characterized as return of capital, ideally paid monthly or more often, because that character defers the tax and leaves more cash to deploy into Bitcoin accumulation or tactical rotation without selling anything.

The framework’s named income-Ballast instrument is STRC, Strategy Inc.’s variable-rate perpetual preferred stock. As of January 2026 it paid 11 percent a year in monthly distributions, historically characterized as return of capital; that characterization is set each tax year and verified on Form 1099-DIV. STRC qualifies as Ballast under the three-of-five rule on durable cash generation (for a preferred, its distribution record: paid in every period since issuance), low factor correlation (the monthly rate reset is designed to hold its price near its $100 stated amount) and a liquidity-resilient valuation (its $100 stated amount and rate-reset support). It carries no agency credit rating as a new instrument; the framework reads the issuer’s Bitcoin asset coverage as its measure of balance-sheet strength.

Update, September 2026. On June 26, 2026, during a Bitcoin selloff, STRC closed at $74.57 (intraday low $71.25). It was back above $84 by June 30 and closed at $99.56 on September 29, close to its 52-week high of $100.42. Distributions were paid throughout; the rate is now 12 percent a year, paid semi-monthly.

The mechanism underneath is deferral. A return-of-capital distribution is not taxed when you receive it; it lowers your cost basis (IRC §301(c)(2)). Once basis reaches zero, further return-of-capital distributions are taxed as capital gain when received (§301(c)(3)). The deferred gain is taxed when you sell, unless a step-up at death resets the basis first. Your broker reports these distributions on Form 1099-DIV, Box 3 (nondividend distributions), and the issuer files Form 8937 describing the basis effect.

The deferral is only as good as the instrument paying it.

Like any partner, the tax code shares losses too, within limits. A realized loss offsets capital gains and up to $3,000 of ordinary income a year ($1,500 if married filing separately), and net losses beyond that carry forward. Return-of-capital distributions steadily lower basis, which makes harvestable losses less likely, not impossible; such a position is held for basis step-up rather than harvesting. The framework’s harvesting candidates are tactical satellites, kept at satellite size. Bitcoin is left off the list: a harvest is a sale, and the accumulate-only rule in Part 3 makes no exception for one. Directly held Bitcoin is not covered by the wash-sale rule under current law. On September 16, 2026 the House Ways and Means Committee advanced H.R. 10357, the Digital Asset Tax Certainty Act, 38 to 5. As approved, it would extend the rule to digital assets sold after September 14, 2026, the day it was introduced, so a loss taken now could still be caught if it is enacted. It is not law.

For long-duration convex positions, pre-tax accounts have four drawbacks. Every dollar that comes out, gains included, is taxed as ordinary income, so a right-tail win never gets the long-term capital gains rate it would have had in taxable; the deduction going in is the other side of that trade (see the caveat under the Tax Wedge). RMDs start at age 73, or at 75 for anyone born in 1960 or later, under current law, and they force taxable withdrawals whether you need the money or not. Heirs generally owe income tax on what they withdraw, with no step-up in basis. And withdrawals before age 59 and a half generally incur a 10 percent additional tax on top of income tax.

The exhibit below splits one balance into both claims; read the deferred tax as a second position riding inside your balance.

Pre-tax accounts still earn their place in three cases.

Everything above rests on rules Congress can rewrite: contribution limits, income thresholds, distribution ages, deduction rules, step-up at death, the wash-sale rule, tax rates, and the tax-free status of Roth withdrawals itself. The framework assumes none is permanent. Its routing usually leaves money in more than one kind of account (the employer match and ordinary-income payers often land in pre-tax), and that spread limits how much damage any single change can do.

The two tax-advantaged accounts are exposed differently. Roth money has already paid tax at a known rate, so an adverse change would be new tax on capital that was taxed once already. Pre-tax money has paid nothing yet, so an adverse change raises or accelerates a bill that was always coming. Roth rules carry no guarantee; the framework favors Roth because, under legislative uncertainty, it has the better risk asymmetry of the two.

This part decides where each kind of holding lives; what to hold, how large, and when to rotate belong to Part 5 and Part 6.