Insurance that burns almost every week and still pays back 2.95 for every dollar

The question

Selling options brings in a small, steady premium on most days and gives a lot of it back in a crash. That's how our Pendulum earns, and that's where it bleeds. So the first question was simple: can cheap puts far below the price stop the bleeding without eating all the premium?

The second question came later, once the first numbers looked better than we expected: do those puts make money on their own, with nothing to insure?

How we tested it

Two years of Deribit data on BTC and ETH. Option prices were rebuilt hour by hour from the exchange's real volatility index (DVOL), then every trade paid the spread we measured on actual quotes plus Deribit's fee. A new put every week, held to expiry.

Entries were spread across four points of the week so no single lucky day decides the result. After a week when the put paid, the next week is skipped: crashes come in clusters, and a second put in a row is bought at the price of fear. About 21 separate experiments in two rounds, each with its code and report.

As insurance: it works

On the option-selling strategy, far puts cut the deepest drawdown from 23% to 15% on BTC and from 30% to 18% on ETH. And the insurance paid for itself: net of its cost, the insured strategy finished the crash cycle $1,890 (BTC) and $3,887 (ETH) better off. Calls as insurance against a sharp rise added nothing useful, crypto's upside tail is too thin to pay for.

Live paper trading at real prices since 11 May shows the same thing. The Pendulum without insurance has slipped from $50,000 to $47,256. The three insured versions sit at $50,585, $54,011 and $59,491 (data as of 25 September 2026). So what separates the red line from the green ones is the insurance that loses money in most weeks.