The Options Wheel (S&P 500)
WHL · The Wheel on SPY
Cash-secured put + covered-call premium harvest (the wheel, SPY)
On paper it looks like it crushes SPY — 31% vs 15% CAGR at half the drawdown — but the entire edge lives in the flat-VIX premium assumption. Halve the modelled credit and it falls to 5.7% CAGR, below SPY. A real-but-fragile premium harvest: in paper-trading, no capital, pending a clean 90-day live forward test before any Approved verdict.
- Category
- Volatility
- Window
- 2021-01-04 → 2025-12-30 (1,254 trading days)
- Instruments
- SPY (30-DTE ~30-delta options, modelled)
- Timeframe
- Monthly (~30-DTE cycles)
- Tested
- 2026-06-29
Pending validation
A scaled equity curve appears here once this strategy clears the data needed to compute one honestly. We don't show a curve we can't stand behind.
See what the gates require →How it works
The bet
What market behavior this strategy is wagering on.
It bets that SPY options chronically price in more volatility than the market actually delivers, so someone who repeatedly sells that insurance — cash-secured puts, then covered calls on any assigned shares — pockets the implied-minus-realized gap as a steady premium.
Technical detail
This is the equity variance-risk premium expressed through the wheel: ~30-delta, 30-DTE short options on the broadest US equity ETF. Documented real-world capture is modest — roughly 1-3% annualized excess over buy-and-hold SPY — and it comes bundled with a fat left tail, not free money.How it decides
What makes it enter, size, and exit a position.
With no position it sells a 30-day put about 30 delta out of the money, then buys it back at 50% profit or at 21 days to expiry, whichever comes first. If the put is in the money at that point it takes the shares and starts selling covered calls the same way; if the shares are called away it goes back to selling puts. One contract per $50,000 of equity — no leverage, no averaging down.
Technical detail
Strikes are solved from Black-Scholes for a ~0.30 target delta; premiums are modelled off a flat VIX-as-IV with the 13-week T-bill (^IRX) as the risk-free rate — free data only, no option chain. Sizing recomputes at each fresh put as floor(equity / $50k), so it compounds with the account but never martingales into a loser (G_RETAIL_SAFE).How it can break
The regimes and failure modes that turn the edge negative.
A fast crash with shares assigned is the kill: the covered call caps any rebound while the stock keeps falling, so you eat the downside with no upside. 2008- and 2020-scale drops hurt this strategy most. And in this backtest almost the entire apparent edge is the modelled premium — cut the assumed credit in half and the wheel's excess over buy-and-hold SPY disappears.
Technical detail
At ~30-delta a cash-secured put is assigned in roughly 30% of cycles (it expires OTM ~70% of the time), and over enough repetitions getting put the stock is a routine event, not a rare one. The conservative run (premium ×0.5, a crude stand-in for skew and real fill quality) drops CAGR from 31.5% to 5.7% — below SPY — while drawdown stays around 18%. The edge is not robust to the premium assumption. Non-IRA accounts also bleed short-term-gains tax on every cycle.
Explainer compiled 2026-06-29 · opus-4.8
Equity curve — vs S&P 500
Growth of $49,977 over 2021-01-04 → 2025-12-30, normalized to a common base. Coded from the committed backtest CSV — not an image.
Source: yfinance (SPY, ^VIX, ^IRX); premiums modelled via Black-Scholes (flat VIX-as-IV). scripts/backtests/wheel_spy.py
Market context
Live chart
Chart powered by TradingView. Live prices can differ from the point-in-time dataset used in the published test.
A premium harvest that may be a mirage
The wheel is the friendliest-looking options income strategy: sell a put, get paid; if you’re assigned, sell calls on the stock, get paid again; rinse and repeat. Run mechanically on SPY over 2021-2025, the modelled version looks spectacular — a 31.5% CAGR against buy-and-hold SPY’s 14.9%, at a shallower drawdown (-11.5% vs -24.5%).
That result is too good, and the backtest is built to show you why.
The numbers — and the haircut that erases the edge
| Scenario | CAGR | Max drawdown | Sharpe | Sortino |
|---|---|---|---|---|
| Wheel — modelled (flat VIX) | 31.5% | -11.5% | 2.90 | 3.85 |
| Wheel — conservative (premium ×0.5) | 5.7% | -18.2% | 0.54 | 0.65 |
| SPY buy & hold (total return) | 14.9% | -24.5% | 0.90 | 1.23 |
The base run sells options at a premium derived from a flat VIX-as-implied-vol with no volatility skew. Real out-of-the-money puts trade at a higher implied vol than that, so a real seller collects more on the put side — but real fills, bid/ask, and skew on the call side all cut the other way, and the biggest single unknown is simply how much premium survives to your account. The conservative column halves the modelled credit as a deliberately blunt stand-in for that uncertainty. Under it the wheel earns 5.7% — below SPY’s 14.9% — while the drawdown barely improves.
That is the finding. The wheel’s apparent outperformance is a bet on premium capture, not a structural advantage over owning the index. Across 196 modelled cycles the base run wins 70% of the time (consistent with selling a ~30-delta put that expires worthless ~70% of the time); the conservative run wins 35%. Same mechanics, opposite verdict — the only thing that changed is the premium.
The decision (value-threshold vs SPY)
Our promotion rule is simple: a strategy has to beat buy-and-hold SPY on a risk-adjusted basis to earn capital. This one straddles the line — it clears the bar comfortably under optimistic pricing and fails it under conservative pricing. When the verdict flips on an unverified assumption, the honest call is not “deploy” and not “reject”; it is paper-trade and let live data break the tie. No capital is committed. (See /framework for the full value-threshold doctrine.)
How the backtest works
- Data (free only): SPY,
^VIX, and^IRXdaily from yfinance, 2021-01-04 → 2025-12-30. - Premiums: Black-Scholes, with a single flat implied vol per day taken from VIX/100 and
^IRX/100 as the risk-free rate. No paid option chain. Strikes are solved for a ~0.30 target delta and rounded to the $1 SPY grid. - Costs: shorts sold at bid (−4%), bought back at ask (+4%), $0.65/contract commission each transaction.
- Benchmark: SPY buy-and-hold uses adjusted close (dividends reinvested); the wheel is not credited dividends while holding assigned shares. Both choices make the comparison harder for the wheel, on purpose.
- Assignment: decided at the 21-DTE management date rather than a true Friday settlement (a simplification; understates pin and overnight-gap risk on the short strike).
Everything is reproducible: python3 scripts/backtests/wheel_spy.py regenerates data/wheel_spy_equity.csv and data/wheel_spy_metrics.json.
What happens next
A forward-test scaffold (scripts/forward_test/wheel_spy_daily.py) now logs the mechanical signal — open CSP / roll / assigned / covered call / called away — against the live SPY close every trading day, starting 2026-06-29 (Day 0). It does not trade; it records what the rules would do. Day 90 is the earliest this can be considered for an Approved verdict, and only if the live path tracks the optimistic scenario rather than the conservative one. Until then it stays PAPER.
Frequently asked
Does the wheel on SPY beat buy-and-hold?
It depends entirely on how much option premium you actually collect. In this 2021-2025 backtest the modelled wheel returned a 31% CAGR vs 15% for buy-and-hold SPY — but that uses an optimistic flat-VIX premium with no skew. Halving the assumed premium (a crude realism haircut) drops the wheel to 5.7% CAGR, below SPY. The honest answer: not reliably, and not until a live forward test confirms which premium scenario reality resembles.
What is the biggest risk of running the wheel?
A sharp crash while you hold assigned shares. The covered call caps your recovery on the way back up while the stock keeps falling, so you absorb the full downside with no upside. Getting assigned and holding the stock through a drawdown is a routine part of the strategy, not a rare accident.
Is this a validated, ready-to-deploy strategy?
No. It is in PAPER status — backtested on free data with modelled premiums, now in a live signal-logging forward test. No capital is committed. It cannot earn an Approved verdict until it clears a clean 90-day live forward test, the 12th gate of the validation framework.
Methodology: Independent research screen — the full 12-gate battery was not run; data and execution limits are stated in the report. Full reproducible report: scripts/backtests/wheel_spy.py in the source repository.Author: Validated Research Team (Methodology v1.0 — 11-gate validation). Backtests are not investment advice.