10-Year Options Strategy Backtest

Underlying. All six scenarios use the identical underlying ("the index", de-identified — a broad U.S. equity market-index ETF), the identical 10-year window (2016-09-12 → 2026-09-11, 2,514 trading days), and the identical data source (Yahoo Finance daily dividend-adjusted closing prices). None are re-anchored to today's date, so every line is directly comparable on one x-axis.

The five prior scenarios (already shown to Vishal, replotted here unchanged):

• VOO Buy & Hold — buy at the start, never trade again. The honest, zero-selection-bias baseline.
• Sector Momentum Rotation — walk-forward, no-lookahead rotation across 11 pre-defined SPDR sector ETFs.
• Trend Following — walk-forward, no-lookahead trend-following overlay on the same sector universe (8 trades over 10 years).
• The Wheel — monthly cash-secured ~30-delta puts; if assigned, switches to monthly ~30-delta covered calls on the resulting shares. Bounded risk throughout.
• Systematic Long Calls (aggressive) — each month, spends a fixed 10% of current equity buying at-the-money 1-month calls. High variance by design; most cycles expire worthless, but the tail compounded to an extreme outcome in this specific historical window — shown here exactly as previously modeled, for comparison, not as a forecast.

The new scenario — Uncovered (Naked) Call Writing. Every ~21 trading days (~1 month), sell ~30-delta out-of-the-money calls on the index without owning any shares (fully naked). Collect the premium up front; if the index closes above the strike at expiry, the seller must buy at the market to deliver — in principle an unlimited-tail loss. This is the only scenario on the chart with uncapped downside, and the only one subject to a margin requirement.

Options pricing (Wheel, Long Calls, and the new Naked Calls scenario). Black-Scholes, European, no early exercise. Volatility input = trailing 21-trading-day realized volatility of the underlying, annualized, fixed at each cycle's entry only (no lookahead). Risk-free rate = constant 3.0%/yr, credited daily on idle cash. Dividend yield q=0 (the series is already dividend-adjusted). Slippage = a flat 1% haircut on every option premium. Target delta = ±0.30 for all three options strategies.

Margin modeling, specific to the new Naked Calls scenario. Because these calls are uncovered, a standard Reg-T margin requirement is computed from the model's current mark every single trading day: max(20%×spot − out-of-the-money amount, 10%×spot) + current option value, per 100 shares. Contracts are sized at each month's entry so that day's margin requirement is fully covered by that day's account equity (100% utilization of modeled Reg-T buying power — a maximally aggressive, no-cushion sizing rule, deliberately chosen so the margin constraint meaningfully shapes the result). If modeled equity ever falls below the day's margin requirement, the position is forced closed (bought back) at that day's model price, realizing the loss immediately, even mid-cycle. A sensitivity check across 50%–100% margin utilization produced the same qualitative result (steady underperformance, no full wipeout, frequent forced closes) in every case — this is not an artifact of one specific sizing choice.

What actually happened in the model, worth knowing: 101 of 120 monthly cycles were margin-called and force-closed before reaching natural expiry — the account was never assigned in-the-money "the normal way" even once. The single worst event was the COVID crash-rebound: a position entered on 2020-03-16, right as realized volatility spiked to ~79% annualized, was forced closed the very next trading day as the index ripped +6.5% in one session, costing roughly 15% of the entire account in a single day. The account never fully reached $0 in this particular 10-year realization, because the daily margin check catches trouble quickly — but this is a real limitation of the model, not a comfort: it only checks the closing price each day, so a large overnight or intraday gap (a short squeeze, a surprise rally, a crash-up) that this specific decade didn't happen to produce at the worst possible moment could still wipe the account in reality. Real brokers also typically impose house margin stricter than bare Reg-T, charge commissions, and face real bid/ask spreads and assignment mechanics — none of which are modeled here, all of which would make a real account's result worse, not better, than what's shown.

Honest limits that apply to every scenario on this page. No real option market data (quotes, bid/ask, actual fills) exists for us to backtest against for free — every options number here is a model, not a historical record. No commissions or taxes are modeled anywhere. Past performance, even modeled past performance, is not predictive of anything. This page exists to make the shape and scale of the risk visible side by side, not to recommend any of these as an actual strategy.