GLD long straddle backtest
the gold fund. Buy an at-the-money call and put together, 30 to 45 days out. It pays when the price moves far in either direction.
What the history shows
GLD long straddles were replayed on 228 entries across 125 months (2016 to 2026), averaging +1.2% a trade on the dollars committed, with a 95% interval of -11.0% to +14.8%. The interval includes zero, so this history cannot separate the result from chance.
Only 41% of trades won, so the result leans on the winners being larger than the losers (a profit factor of 0.92). The worst single trade lost 100% and the account's deepest fall was 62% (with 10% of the account at risk each month).
The last year's Sharpe ratio was 1.00 against 0.09 over the 5-year window (one year is only 12 monthly readings, so it is a rough guide), so recent conditions have favored this setup more than its longer history did.
Charging an extra $0.01 a share on every option leg moves the average trade from +1.2% to +0.9%; at $0.05 it is -0.5%. That is a 156% drop between the two, so the result is sensitive to execution: it needs liquid contracts and limit orders near the midpoint. The average return reaches zero at about $0.04 a share per leg.
The best entry year was 2016 (+26.5% a trade) and the worst was 2021 (-22.8%). 52% of the entries come from before March 2022, when prices are estimated from daily closes with a haircut measured against real quotes; the rest use real bid and ask quotes. These are historical replays, not live quotes and not a forecast.
Out-of-sample windows
| Window | Sharpe | Sortino | Max drawdown | Win rate | Profit factor | Per trade | Trades |
|---|---|---|---|---|---|---|---|
| Last 1 year | 1.00 | 2.40 | 9% | 52% | 1.41 | +19.4% | 25 |
| Last 3 years | 0.66 | 1.48 | 30% | 45% | 1.19 | +15.6% | 73 |
| Last 5 years | 0.09 | 0.16 | 40% | 41% | 0.99 | +2.6% | 118 |
| Full history | 0.04 | 0.06 | 62% | 41% | 0.92 | +1.2% | 228 |
Slippage and commission sensitivity
| Fills | Net return per trade | Change |
|---|---|---|
| Replay fills (shorts at the bid, longs at the ask) | +1.2% | n/a |
| Extra $0.01 a share on every leg | +0.9% | -0.3% |
| Extra $0.05 a share on every leg | -0.5% | -1.7% |
Fills in the replay already sell at the bid and buy at the ask, and a commission of $0.65 a leg is included. The extra slippage rows add to that.
Year by year
| Year | Entries | Average per trade |
|---|---|---|
| 2016 | 15 | +26.5% |
| 2017 | 22 | -10.7% |
| 2018 | 18 | -15.9% |
| 2019 | 23 | +15.6% |
| 2020 | 19 | -15.2% |
| 2021 | 20 | -22.8% |
| 2022 | 22 | -10.9% |
| 2023 | 24 | -0.5% |
| 2024 | 24 | +19.6% |
| 2025 | 24 | +24.8% |
| 2026 | 17 | -1.6% |
Change the delta, days to expiry or wing and replay your own version on GLD.
Common questions
What is the historical win rate of GLD long straddles?
41% of 228 replayed entries made money (2016 to 2026). The average trade returned +1.2% on the dollars committed and the worst lost 100%.
How much does slippage change GLD long straddles?
An extra $0.01 a share per leg moves the average trade from +1.2% to +0.9%, and $0.05 moves it to -0.5%. The average return reaches zero at about $0.04 a share per leg.
How has GLD long straddles done recently?
Trailing windows ending 2026-09-03: 1 year: Sharpe 1.00, win rate 52%; 3 years: Sharpe 0.66, win rate 45%; 5 years: Sharpe 0.09, win rate 41%.
What is the biggest risk of GLD long straddles?
The worst single trade lost 100% of the dollars committed, and the account's deepest fall from a peak was 62% (with 10% of the account at risk each month). The result depends on winners outweighing losers.
Method
How these numbers were made: each strategy was priced on weekly or fortnightly entry days from January 2016 to September 2026 on real option quotes (shorts at the bid, longs at the ask, fees per leg), 30 to 45 days to expiry, about 25 delta, wings 5% away, held to expiry and settled on the underlying's close. Entries before March 2022 use daily closes with a haircut measured against real quotes. Returns are profit divided by the dollars committed. Sharpe and Sortino use one observation per entry month and are annualized with the square root of 12, with no risk-free rate. Max drawdown compounds each month's average return at the strategy's usual sizing: the full cash set aside for a cash-secured put or covered call, 10% of the account at risk each month for spreads, condors and long options. The rules have no fitted parameter, so every window is out of sample. Intervals are a month-clustered bootstrap. Slippage is charged on top of fills that already cross the spread. This is education, not a recommendation to trade anything.