Price data
Daily closing prices come from stockanalysis.com, adjusted for splits and dividends, with older history from Yahoo Finance, going back up to 10 years (less for newer listings). The data is refreshed regularly; the current data runs through 2026-10-16.
Option prices are modeled, not quoted
There are no historical option quotes on this site. Every call and put is priced with the Black–Scholes formula using the stock price that day, the strike, the time to expiry, a risk-free rate (by default that month's 3-month T-bill rate, from the Federal Reserve's FRED database), and a volatility estimated from the stock's own past prices. Only data available on that day is used — no look-ahead.
Volatility estimators
- Market-like (default): an equal blend (in variance) of realized volatility over a window as long as the option and over the past year. Real implied volatility stays anchored to longer-run levels — it doesn't collapse in a quiet month — and this blend tracks that much better than a short window alone.
- Tenor-matched: realized volatility over a trailing window as long as the option only, so a 90-day call uses ~90 days of history.
- 21-day: the classic one-month historical volatility (HV20), regardless of the option's length.
- EWMA: RiskMetrics exponentially weighted volatility (λ = 0.94), which reacts faster to recent moves.
Volatility premium and haircut
Implied volatility in real option markets has historically run above the volatility that actually followed — the "variance risk premium" option sellers earn. The vol premium setting scales the historical estimate up to approximate this. Its default, 1.05×, is calibrated against real funds: XYLD and QYLD sell at-the-money monthly calls on the S&P 500 and Nasdaq 100, and with 1.05× a modeled at-the-money buy-write on SPY and QQQ — averaged over every possible start date — returns what those funds actually returned since late 2021 plus their ~0.6% annual fees. It is an average — on any given stock or month, real implied volatility can sit well above or below it. Options are also priced with skew, as in real equity option markets: a put's volatility is raised the further below the price it is struck (about 15% richer at 10% out of the money), and a call's is lowered the further above (about 7% lower for a 0.30-delta call, flattening at 9% lower further out, where real single-stock options tend to curve back up). Without call skew the model overprices exactly the out-of-the-money calls covered-call sellers write. On SPY, this model's 0.30-delta covered call returned 14.9% a year against 15.4% for holding. Two cautions: the 1.05× calibration was fitted on the same period it is then compared with, so it is a fit, not an out-of-sample test; and volatility measured from past prices runs high right after a crash, so the model overprices calls then — real buy-write indexes lagged the S&P 500 badly in 2020, where this model has the calls ahead. The premium haircut (default 5%) then trims each premium to stand in for bid/ask spreads and fills worse than the mid.
Covered call rules
Each cycle sells one call. By default the strike is set by delta — a 0.30-delta call is the strike the model gives a roughly 30% chance of finishing in the money — recomputed every cycle from that day's volatility, so strikes sit further out when the stock is jumpy. You can switch to a fixed % above the price instead. At expiry you keep the premium; if the stock finished above the strike you either pay the difference (buy the call back at intrinsic value, keep the shares and write the next call — the default) or, with let the shares go, are assigned and hold the cash, earning the risk-free rate, from then on. Income per year, buyback cost and how often calls finished in the money use every trading day as a start date (overlapping cycles). The equity curve, and every annual return built on it, reinvests each cycle's option profit or loss in the shares (like Cboe's BXM index) and is averaged over every possible start day in the first cycle, so the result doesn't hinge on one lucky or unlucky starting date.
That is why net from calls and covered call vs. holding are different numbers. Net from calls is a simple average of what each call added or cost, as a share of the stock price. The gap versus holding is compounded, and capping the upside also smooths the ride: a steadier path loses less to volatility drag, so on a volatile stock the compounded gap can be several points a year better (or worse) than the net.
Wheel rules
Start fully in cash. Sell a cash-secured put; if it finishes in the money you are assigned the shares at the strike. Then sell covered calls until the shares are called away, and return to selling puts. Premium is added to collateral, so the position compounds. Idle cash earns the risk-free rate by default. Optionally, calls are never written below your cost basis.
What the backtest ignores
- Early assignment, including ahead of ex-dividend dates.
- Taxes, commissions, and margin requirements.
- Real bid/ask spreads and liquidity beyond the haircut — thinly traded options can be much worse.
- Earnings and stock-specific skew: implied volatility jumps ahead of earnings are not modeled, and every stock gets the same skew shape, so premiums on names with big event risk are likely understated.
- Real implied volatility: the model estimates it from past prices; on any given day the market can price options well above or below that.
Because of this, the site is best for comparing strategies and stocks with each other, not for predicting the exact income you would have earned. Nothing here is investment advice.