Does Options Automation Match the Backtest? 8 Weeks of Live QQQ Trades
We automated a QQQ double calendar and compared eight weeks of live trades with the backtest, looking at strike selection, fills, and exit timing.
This eight-week exercise compared automation with a historical backtest. It was not an assessment of the strategy’s ongoing profitability. For a later discussion of W/F DC pricing and performance, see our companion video.
Matt from Option Omega automated a QQQ double calendar and compared eight weeks of live trades with the backtest. Strike selection matched in all eight weeks. Over that window, the live trades made $84 on one lot, compared with about $50 in the backtest.
We’ll walk through the backtest, automation settings, and week-by-week results to see where they matched and where they differed. This was an eight-week comparison of one strategy. Matt tweaked the exit time to be a few minutes before major news.
The backtest: a QQQ 1-3 double calendar
The strategy came from a backtest built on the OO Show, Option Omega’s daily morning YouTube show. It is a “13” double calendar on QQQ: short options one day from expiration, long options three days out, with a put calendar below the market and a call calendar above it.
| Setting | Value |
|---|---|
| Underlying | QQQ |
| Structure | 1-3 double calendar (13DC) |
| Frequency | One entry per week, shortly after the open |
| Exit | Time exit in the afternoon of the short legs’ expiration day, plus leg delta exits |
| Profit target | None |
| Size | One lot |
| Commissions | About 15 cents, to match a lower-cost broker |
| Slippage | Three cents on entry and three cents on exit |
| Blackout dates | Holiday weeks |
Before automating it, Matt checked a few changes in the backtester:
- A tent-sag exit. He tried exits based on the short/long price ratio. A 0.05 threshold lowered the capture rate, and lower thresholds barely ever triggered, so he kept the original version.
- Profit targets. Matt tested profit targets with two-hit confirmation and capped profits. Those variations reduced performance in the backtest he examined, so he retained the version without a profit target.
- News-time prints. Some trades showed brief profit spikes around scheduled announcements. Matt viewed those marks cautiously and avoided those specific entry and exit minutes in his testing.
The automation settings
Matt transferred the saved backtest into automation, selected his account, and set the position to one lot. He then configured the following order-handling settings for this comparison.
- Entry: 1-cent price increments, a 3-minute entry window, and a 1-cent starting offset. The offset set the initial order price; the eventual fill determined how it compared with the backtest’s 3-cent slippage assumption.
- Normal exit, or time exit: a more relaxed exit, stepping every nine seconds.
- Above-delta exit: no offset, stepping every four seconds.
- Below-delta exit: stepping every six seconds.
- Exact DTE and exact strike offset, so the trade stays a true calendar rather than becoming a diagonal.
- Holiday blackout dates, added as the holidays came up.
Week-by-week: backtest vs. live
Here is how each trade compared. The strikes listed are the put and call strikes chosen.
| Week | Strikes (live = backtest) | What happened |
|---|---|---|
| 1 | 607 / 617, match | Entered for $3.12 and exited for $2.60 on a big, sharp candle, about a 17% loss. Fills were identical to the backtest. |
| 2 | 625 / 639, match | Live entry was recorded at $3.28 versus $3.26 in the backtest. The trade was a winner, with the exit time moved a few minutes earlier ahead of FOMC. |
| 3 | 614 / 631, match | A loser. The backtest was one cent worse on entry and three cents worse on exit than live. |
| 4 | 616 / 628, match | Live beat the backtest by a penny on entry and two cents on exit. |
| 5 | 590 / 613, match | The backtest made $71; live made $73. Exited one minute early because of news. |
| 6 | 615 / 627, match | Entry took seven attempts and filled about three cents worse than the backtest. Lost almost a dollar. |
| 7 | 616 / 630, match | The backtest lost $29; live lost $24. |
| 8 | 603 / 618, match | Live made 15.5%; the backtest made 15.4%. Exited on the above-delta, or downside, exit in the early afternoon. |
After four weeks, Matt reran the backtest over those trades using the actual fees. It showed a $12 loss, while the live account was up $5.
Thanksgiving week was skipped on purpose, since the strategy does not trade short holiday weeks.
Monitoring a live calendar with the modeling tool
During week two, Matt opened the trade in the modeling tool from the strategy page. The model refreshes every few seconds, separate from the automation platform’s once-a-minute check.
With FOMC the next day, he used it to:
- Examine the model’s projected profit range—roughly 625 to 640 on QQQ in that example—and how it changed with time and volatility.
- Compare implied volatility by expiration. The 1-day and 3-day IVs were nearly the same in this example, and Matt used the model to examine how changes in either could affect the position.
- Model what-ifs: both IVs collapsing after FOMC with the 3-day holding a little more, or both rising.
- Add Greeks such as delta to the chart, including a 3D view.
Matt finds the model especially useful once a trade is open. In this exercise, he used it alongside the backtest and live results to examine how the position behaved.
Matt found more value in exploring changes in time, price, and volatility than in simply watching the position refresh.
Managing the live trade and checking it against the backtest
Two kinds of adjustments came up.
FOMC exits. On FOMC days, Matt moved the trade’s exit to 2:27 so it would close a few minutes before the Fed press conference, without changing the saved strategy. His rule of thumb is to avoid entering or exiting at news times because prices can be “blippy.”
Should I take profits? In week two, the trade was up 44%, and Matt was tempted to close it. Instead of guessing, he ran the backtest uncapped, then with profit targets.
| Profit target | Result vs. original |
|---|---|
| None (original) | MAR 3.7 |
| 40% | Worse |
| 50% | Not better |
| 75% | Same as original |
None of the profit targets he tested improved the historical comparison, so Matt left the existing exit rules in place. That explained his decision at the time; it did not establish how the trade would perform afterward.
The results
| Measure | Backtest | Live |
|---|---|---|
| Strike selection | — | Matched in all eight weeks |
| Profit on one lot | About $50 | $84 |
Matt explained the gap. The backtest assumes three cents of slippage each way, which turned out to be more than the live fills usually cost. A couple of exits came a minute or a few minutes early around news, which also nudged results. His philosophy is to err on the side of too much slippage: “realistic but conservative.”
He was candid that the strategy itself was not spectacular in this window. Its capture rate was 3.2%, and the equity curve had chopped for about six months.
At the time of the video, Matt described a backtest covering roughly three years, including flat stretches of about four months in 2023 and eight months beginning later that year. Those were the historical results available then.
At the end of the recording, Matt added the remaining holiday weeks to the blackout dates and planned to continue the comparison.
Key takeaways
- Strike selection matched in all eight weeks. That was one part of the comparison; fills and exit timing also affected the results.
- Slippage assumptions affected the comparison. The backtest included three cents on entry and exit, which exceeded the slippage on most observed fills. Live finished ahead in this eight-week window: $84 versus about $50 (very small sample size!).
- One entry took seven attempts and filled a few cents worse than the backtest. The comparison included execution differences in both directions.
- Matt checked profit-target variations before changing the live trade. None improved the historical results in that comparison, so he kept the existing exit rules.
- Matt tweaked the exit time to be a few minutes before major news. Those timing changes matter when comparing the live trades with the saved backtest.
What this comparison tells us
The point of this exercise was to compare automation with the backtest.
Watch the full video: Does Options Automation Actually Work? Straightforward Setup + Live Trades (Option Omega).
Backtest results are hypothetical and do not guarantee future performance. Options trading involves substantial risk.
