RET-CC-002 — Covered Calls After Tax

This page explains what the Tax Lab does, what it assumes, and where it should not be trusted. Read it before you draw a conclusion from the numbers.

Why this study exists

RET-CC-001 compared a buy-and-hold retirement with one running a systematic 30-delta covered-call overlay. It found that the overlay adds no return — $8.60M of gross premium netted to −$314k after assignments — and improves retirement resilience anyway, lifting the 30-year success rate from 93.0% to 95.7%.

Its stated single largest limitation was that taxes were not modelled. Since the net option contribution was already slightly negative, taxes could plausibly flip the conclusion. This study tests exactly that.

The one thing that changed

RET-CC-001 settled each call at intrinsic: when a call finished in the money, the portfolio simply paid out the difference. No shares changed hands. Before tax that is exactly equivalent to being assigned and buying the shares straight back, which is why it was a perfectly good simplification there.

After tax it is nothing like equivalent. Assignment sells your shares. The covered position is sold at the strike, the gain is realised, and the shares are repurchased at market with a fresh cost basis. At 100% coverage and a 37.8% assignment rate, that happens roughly every two and a half months for thirty years, and the portfolio never accumulates a deferred gain at all.

Buy-and-hold, meanwhile, defers. In the base case it ends year 30 with 78% of its value still unrealised. The covered-call portfolio ends with essentially none.

How the options are taxed

Two regimes, and the lab lets you switch between them because they give different answers on identical economics.

• SPY equity options (the default). Options on an ETF are equity options.
  • A call that expires worthless: the premium is short-term capital gain, taxed at ordinary rates.
  • A call that is assigned: under §1234(a) the premium is folded into the sale proceeds of the stock, and the character follows the stock. Because a 30-delta out-of-the-money call is a qualified covered call under §1092(c)(4), the stock's long-term holding period is not suspended, so the gain is long-term.
• Broad-based index options (§1256). Cash-settled, marked to market, and taxed 60% long-term / 40% short-term regardless of holding period. Critically, the shares are never called away, so the deferral survives.

Where the effect actually comes from

It is tempting to assume the cost is the rate on the premium. It mostly is not. Two things matter more.

Deferral. A dollar of gain taxed in year 3 and a dollar taxed in year 30 are not the same dollar. Thirty years of compounding on the government's share is the larger part of the gap.

Bracket stacking. Ordinary income sits underneath long-term gains and qualified dividends, pushing them up through the preferential brackets. A retired couple drawing 4% from a $1,000,000 portfolio, with the standard deduction and the senior additions, can sit in the 0% long-term bracket for most of a retirement and pay essentially no federal tax at all. Add short-term premium underneath, and gains that were free start costing 15%. The premium's own tax bill understates its true cost, because part of the damage lands on income the retiree already had.

The tax model

2026 federal, married filing jointly, both spouses 65 or over, standard deduction. Constants live in tax.js and are sourced to IRS Rev. Proc. 2025-32.

• Seven ordinary brackets, 10% through 37%.
• Three preferential brackets (0/15/20%) for long-term gains and qualified dividends, stacked on top of ordinary taxable income.
• Standard deduction $32,200, plus $1,650 per spouse aged 65+, plus the OBBB senior deduction of $6,000 per qualifying taxpayer with its 6% phase-out above $150,000 — expiring after 2028, as enacted.
• Net investment income tax of 3.8% above $250,000 of MAGI. This threshold is deliberately not indexed in the model, because it is not indexed in the statute and has not moved since 2013. Over thirty years that is a real and growing tax.
• Capital-loss netting by character, the $3,000 annual offset against ordinary income, and carryforward of the rest.
• Every indexed threshold is inflated along the same stochastic inflation path as the withdrawal, so bracket creep is not quietly smuggled in as a finding.
• The year's bill is paid out of the portfolio the following January, which is when a retiree actually writes the cheque.

Cost basis

Each portfolio carries an average-cost basis pool. Withdrawals sell shares and realise long-term gain pro rata; dividends are taxed as qualified dividends when paid and reinvested, raising basis.

Lot-level FIFO would be more faithful, but 10,000 paths × 360 months × up to 360 open lots is on the order of a billion lot operations, which is not something a browser should be asked to do while you wait. In a pool that is appreciating on average, average cost realises less gain than FIFO would — so this choice understates the tax drag. The bias runs against the study's finding, not toward it.

Comparing terminal wealth honestly

The headline wealth number is terminal value after liquidation tax: the ending balance less the tax still owed on everything unrealised inside it. Comparing raw statement balances would flatter whichever portfolio deferred more, which is precisely the portfolio under test — so it would build the answer into the measurement.

Both figures are on the results table. The honest range sits between them, and it is worth knowing which way each errs: a step-up in basis at death would erase buy-and-hold's deferred gain entirely, which makes the after-liquidation figure the conservative one for this study's conclusion.

What is not modelled

• State income tax. Federal only.
• Rolling the call. The obvious mitigation, and a real one — rolling defers the stock sale instead of triggering it. This study takes assignment as it comes, which is the strict reading of the RET-CC-001 protocol. Treat the headline number as the cost of not managing assignment.
• Required minimum distributions in the traditional IRA case.
• Social Security provisional-income rules. Enter the taxable portion directly in the "other taxable income" field.
• Step-up in basis at death — see above.
• Wash sales, which would rarely bind here because gains dominate.
• Everything RET-CC-001 did not model: early assignment, dividend-capture assignment risk, bid/ask beyond a flat 3% haircut on premium, and any skill in strike selection or timing.

The control that makes this trustworthy

Set Account type to Roth IRA and Other taxable income to 0. No tax is levied anywhere, and the engine reproduces RET-CC-001 to the dollar — same medians, same success rates, same 37.8% assignment rate, same calibration in section 9 of the results table.

That is not a coincidence; it is a regression test that runs in your browser. If the Roth setting matches RET-CC-001, then every difference you see on the taxable setting is attributable to tax and to nothing else. Section 8 of the results table makes the same point from the other direction: the pre-tax strategy CAGR is unchanged at 9.85% versus 9.84%.

Reproducing the published numbers

Seed 20260822, 10,000 paths, normal market, taxable brokerage, SPY equity options, $45,000 of other taxable income, α = 6.0, β = −1.1. About 1.7 seconds. Use Export results JSON for the full result object.