Short Strangle Backtest: 60 Cycles, Undefined Risk
Selling 16-delta puts and calls on SPY and liquid single names. Win rate by delta, the regime where undefined risk bites, and how the strangle stacks up against its defined-risk cousin, the iron condor.
Simulated data for display. Illustrative narrative — not a verified live backtest. Build real backtests on the strategy builder.
All strategy backtests →The Test Setup
Underlyings: SPY plus a handful of liquid single names, entered when IV rank was above 35.
Structure: short strangle — sell an OTM put and an OTM call, no protective wings (undefined risk on both sides).
Entry: 45 DTE, both short strikes at 16 delta (~1 standard deviation).
Management tested: (1) hold to expiration; (2) close at 50% max profit; (3) close at 50% + roll the tested side + 2x-credit stop.
Sizing: one strangle per ~$10k of buying power; single position at a time.
Period: January 2020 – December 2024, sampled to 60 cycles.
Win Rate by Short-Strike Delta
| Delta (each side) | Win Rate | Avg Credit | Avg Winner | Avg Loser | Net / Cycle |
|---|---|---|---|---|---|
| 10 | 78% | $190 | +$95 | -$640 | +$34 |
| 16 ★ | 68% | $310 | +$155 | -$560 | +$62 |
| 25 | 58% | $460 | +$230 | -$520 | +$15 |
| 30 | 52% | $560 | +$280 | -$500 | -$18 |
16-delta is the balance point. Lower deltas win more but the average loser is enormous relative to the thin credit; higher deltas collect richer premium but get tested so often that undefined risk turns negative.
Strangle vs Iron Condor: More Credit, More Drawdown
| Structure | Win Rate | Net / Cycle | Max Drawdown | Risk Profile |
|---|---|---|---|---|
| Short strangle (16Δ) | 68% | +$62 | -24% | Undefined |
| Iron condor (16Δ, $5 wings) | 70% | +$30 | -18% | Defined |
The strangle earns roughly double the net per cycle by skipping the wings — but pays for it with a deeper drawdown and uncapped tail risk. For accounts that can't absorb an outsized loss, the iron condor backtest is the defined-risk version of the same idea.
Performance by Market Regime
| Regime | Years | Win Rate | Net P&L |
|---|---|---|---|
| Range-bound / falling IV | 2021, 2023, 2024 | 79% | +$3,180 |
| Choppy | parts of 2020, 2024 | 66% | +$540 |
| Vol expansion / trending | 2022, March 2020 | 44% | -$1,200 |
Short strangles harvest premium beautifully in calm, range-bound, falling-IV markets and bleed in sustained volatility expansions. The entire edge depends on not getting wiped out in the bad regime — which is what management is for.
Anatomy of the Worst Cycle (2022 Drawdown)
A 16-delta SPY strangle sold for $3.10 credit in mid-2022. A sharp leg down ran straight through the short put as IV expanded, so both the directional move and the rising vega worked against the position at once. Held to expiration it realized roughly -$580. Closing at the 2x-credit stop and rolling the untested call side down recovered about $200 of that.
No protective wing means no automatic floor. In a vol expansion the only floor is the one you set yourself with a stop and a roll — the strategy's survival depends on it.
Five Takeaways
- Sell both sides at 16-delta. The balance of credit and test frequency. 25-delta and beyond turn undefined risk into a liability.
- Only sell into high IV rank. Above 40-50. You are paid for volatility contraction; selling into low IV is selling cheap insurance.
- Always run a stop. Undefined risk means a hard 2x-credit stop on the tested side is mandatory, not optional.
- Roll the untested side. When one side is tested, roll the other closer for extra credit to reduce net risk — a partial conversion toward an iron fly.
- Use the condor if you can't take the tail. The defined-risk version gives up some credit for a hard cap. Match the structure to your account, not your ego.
Price the strangle before you sell it
Check IV rank, then map the breakevens and required move on the calculator before committing buying power.
Related Reading
Backtest narrative is illustrative — built from typical short strangle mechanics and historical volatility regimes, not from live broker fills. Past performance, simulated or real, does not predict future results. See methodology.