Help us double down on what's working, instead of guessing. Takes 5 seconds, totally optional.
research · 4 min read
When to Exit Short Options? Nine Daily-Close Exit Rules Tested on 11 Symbols, 2018 to 2026
We tested nine exit rules checked once a day at the close (take profit, stop loss, time and delta) on 481,000 short-option trades from 2018 to 2026. None beat holding to expiry at the same risk, and on average each one lowered the Sharpe ratio.
Short answer: none of nine daily-close exit rules beat holding to expiry at the same risk. Each rule was checked once a day at 15:55 ET, on 481,467 short-option trades from 2018 to 2026. On average each one lowered the Sharpe ratio: by 0.07 to 0.42 on eight confirmation symbols, where selling every session lost money at these fills, and by 0.03 to 0.45 on SPY, QQQ and IWM, where it made money. Most of the dollar drawdown an exit saved is what a smaller position would also save. Take-profit exits did better than random exits of the same length. Stops and delta exits had a lower Sharpe ratio than random exits.
0 of 9daily-close exit rules that beat holding to expiry at the same risk on the eight confirmation symbols
−0.07 to −0.42average change in Sharpe ratio from adding an exit on the confirmation symbols, at halfway fills
2.9xtypical loss booked by a daily-close stop set at 2x the credit. In a typical combination, one in ten lost more than 4.7x.
30%median share of the COVID loss that came from positions opened before the selloff
Read the numbers with this in mind. On the eight confirmation symbols, selling every session lost money at these fills: the average Sharpe ratio of holding to expiry was −0.21 (+0.08 at midpoint fills), and every symbol was negative. Two descriptive checks: in the 94 confirmation and 86 SPY, QQQ and IWM combinations where holding made money, the average Sharpe change was still negative for all nine exits, and at midpoint fills it was negative for all nine on both sets. The first check was added after the results were read and selects on the outcome of holding. Stops watched during the day were not tested.
Enterprise members
This section is for Enterprise members
Enterprise members get the full results: every exit against holding to expiry and against random exits, what daily-close stops really lose, each exit through six selloffs, results by trade type and maturity, the stability checks and the method.
In this test it lowered risk-adjusted returns on average. The rule was checked once a day at 15:55 ET. On eight confirmation symbols, taking profit at 50% closed 82% of trades early, cut the dollar drawdown by 13% and lowered the monthly Sharpe ratio by 0.14 on average. Scaled to the same volatility, its drawdown was 10% deeper than holding to expiry. It raised the Sharpe ratio in 24% of the combinations. On SPY, QQQ and IWM it kept 60% of the profit.
Do stop losses protect a short option position in a crash?
A daily-close stop cut the dollar loss, but not at the level set. A stop at 2x the credit, checked at 15:55 ET, booked a typical loss of 2.9x the credit, and in a typical combination one stop in ten lost more than 4.7x. In the COVID selloff it kept 56% of the loss of holding to expiry. Over the full period it lowered the Sharpe ratio by 0.15 on average and had a lower Sharpe ratio than random exits of the same length. Stops watched during the day were not tested.
Which exit rule was best for short premium?
None of the nine daily-close rules beat holding to expiry at the same risk. On average each one lowered the Sharpe ratio, by 0.07 to 0.42 on the eight confirmation symbols and by 0.03 to 0.45 on SPY, QQQ and IWM, at fills halfway to the bid or ask. The average change was also negative for all nine at midpoint fills.
Why do exits look good in many backtests?
An exit shortens the trade, so the position is smaller on average and the dollar drawdown falls. Most of that saving is what a smaller position would also give. The fair comparison is at the same risk, and against random exits of the same length.
How much of a crash loss came from positions that were already open?
In the COVID selloff, a median 30% of the loss from holding came from positions opened before 20 February 2020 on the confirmation symbols (26% of the dollars), and 33% on SPY, QQQ and IWM (36% of the dollars). The rest came from positions opened during the fall. The share was higher for 1-month trades than for 1-week trades.
