Covered calls and cash-secured puts are the retail form of selling volatility. The seller collects a premium every month; time decay works in their favor every day. Sell when implied volatility rank is high, so the premium is rich, and buy it back once implied volatility reverts. The buyer on the other side pays for upside or for protection.
volatility risk premium / theta
Options premium selling (covered calls + cash-secured puts)
The idea
How it was tested
Pre-registered on 2026-07-27, with six gates and a projected 16-20% a year. The signal: implied volatility rank above 70 over a trailing 252 days. The structure: covered calls 5-10% out of the money and cash-secured puts 2-5% out of the money, 30-45 days to expiry, on about 20 liquid US large caps. Exit at 50% of max profit, when implied volatility rank drops below 50, or with 5 days to expiry.
Where it died
Four days later, on 2026-07-31, it was recorded as validated: 199.5% annualized in-sample and 357.6% out-of-sample, with no alpha gate. On that record it went into the demo.
The numbers had no measurement behind them. The lab holds no historical option chains, so no backtest on real option prices could have produced them, and the Sharpe ratio and win rate were listed as expected rather than measured. The demo did not fill the gap: its executor priced premiums with Black-Scholes and sent no orders. On 2026-10-17 the strategy was declared unverifiable and withdrawn, and the verdict became void.
The record
Thesis
Sell covered calls and cash-secured puts on liquid large caps when implied volatility rank is above 70: time decay works for the seller every day, and high implied volatility tends to revert, so the premium is collected cheaply and bought back cheaper.
Specification
- signal
- IV rank > 70 (trailing 252 days)
- structure
- covered calls 5-10% out of the money and cash-secured puts 2-5% out of the money, 30-45 days to expiry
- exit
- 50% of max profit, IV rank < 50, or 5 days to expiry
- universe
- about 20 liquid US large caps
Gates
Gate 7 — Jensen alphanot evaluated
After regressing out market exposure, is there alpha left?
Never measured. The lab has no historical option chains, so no backtest on real option prices was possible.
Why it died
Never actually tested. The "validated" version reported 199.5% in-sample and 357.6% out-of-sample annualized, but those figures did not come from a backtest on real option prices; Sharpe and win rate were listed as expected, not measured, and there was no Gate 7 or 8. In the demo, the executor priced premiums with Black-Scholes and sent no orders, so its results were not evidence either. Without historical option chains the strategy cannot be verified, and it was withdrawn.
What it taught
An out-of-sample number higher than the in-sample one was read as "no overfitting" when it was a sign that nothing real had been measured. When the data to price the instrument does not exist, the honest verdict is "unverifiable", not a projected return.
Verdict history
- 2026-07-27Pre-registered with six gates; projected 16-20% a year
- 2026-07-31Recorded as validated: 199.5% / 357.6% annualized, no alpha gate
- 2026-10-17Declared unverifiable: no historical option chains; demo executor never sent orders Void