Covered call
—
annual return
Does selling covered calls beat just holding the stock?
It depends almost entirely on one number that nobody can observe, and this page lets you move it.
The comparison is 100 shares plus a call sold every month at 0.25 delta, 30 days to expiry, against the same 100 shares held on their own. Both run for a year across thousands of simulated markets built from real daily returns. Every other setting is fixed, and they are listed at the bottom of the page. The benchmark here is full ownership of the shares, which is the stricter of the two comparisons — see what this page is not below.
Nobody knows this number for your stock, on the day you sell. That is the problem. Most covered-call arguments quietly assume it is generous.
%
0% — fair value20% — richly priced
Loading…
Difference
—
covered call minus buy & hold
Just holding
—
annual return
Won how often
—
share of simulated markets
This page answers one question at settings somebody else chose.
The lab is where you choose them — your delta, your days to expiry, your horizon, your capital — and, more to the point, where you attack the answer instead of reading it.
- An exposure-matched control, which compares the covered call against the same amount of market exposure rather than against full ownership — the comparison this page cannot make
- Six market regimes swept in one run, each with its own calibration
- A holdout on fresh paths, so a result has to survive a sample it has not seen
- A falsification run at fair pricing, built to kill the result rather than confirm it
- A robustness grid around whatever threshold you picked, because a finding that only works at one setting is not a finding
- Cash-secured puts, the Wheel, LEAPS, PMCC, RMD and household tax, and the rest of the labs
- Results run on the server and saved with the engine fingerprint that produced them, so you can tell later which version answered you
The protocol behind these numbers
PMCC-001 states its question, freezes its method before any result exists, and publishes what it found — including the part where removing one assumption took it from 92 of 96 runs down to 59. Send it to me.
One email with the protocol, and the occasional note when a new study is finished. Nothing else, and no list is sold or shared.
What this page is not
This is a simulation at fixed settings, not a forecast and not a backtest of real option chains. It describes how a model behaves under the assumptions listed below, and changing those assumptions changes the answer — which is the entire point of the dial.
The benchmark is full ownership, not matched exposure. Selling a call caps the upside, so a covered call carries less equity exposure than the shares it is measured against here. In a market that drifts upward, holding less exposure earns less, so comparing against full ownership is the harder test. The lab’s exposure-matched control — which scales the benchmark down to the exposure the strategy actually carried — would generally show the covered call doing better than it does on this page, not worse. Nothing here is weighted against selling calls.
A mean is a thin summary of a strategy that cuts off its own right tail. Covered calls give up the largest gains in exchange for a steady premium, which is why the win rate and the worst-case figure are shown next to the average rather than underneath it. Judging this by the average alone would miss the shape of what you are buying.
Portfolio Research Lab has no brokerage connection, places no trades and handles no real money. Nothing here is investment advice, a recommendation, or a solicitation to trade. Mark is not a registered investment adviser or broker-dealer. Options carry substantial risk of loss and are not suitable for every investor.
Across the research completed so far, results have more often been unfavourable to option sellers than favourable. That is the finding, not a disclaimer.