Feedback on Opening Range Breakout Strategy
By Dave Mabe
Here's a reader question from Brad T. about his opening range breakout strategy and where to go next with it. (Name used with permission, lightly edited for clarity.)
Brad T.
I've been working on an intraday strategy in AmiBroker called ORBreakout v1.6, and I'd value a sanity check from someone who knows this space.
Long only intraday on a curated universe of 564 US-listed symbols (price at least $3, avg dollar volume at least $2M). Each day must gap up at least 2%. Two stacked entries on 12-minute bars. Logic A is a breakout above the 9:30 to 10:00 opening range high with a 1.5x volume confirmation. Logic B is a retest bounce on the premarket high, VWAP, or OR low, gated to fire only after the OR high has been broken once. Hard stop at OR low or pullback low. EMA21 trailing stop. Force flatten at 15:30. Risk-based sizing at 3% per trade, max 8 open. IBKR Pro tiered commissions ($0.0035 per share) are baked into the backtest.
Three things I'd love your push back on:
1. When I split the data into in-sample (2024 H1 to 2025 H1) and out-of-sample (2025 H2 to today), the OOS half came in stronger than the IS half. Is that a real signal of edge, or could I be missing some subtle look-ahead bias?
2. I found that 12-minute bars produced materially better results than both 10 and 15-minute bars (PF 1.25 vs 1.16 and 1.14). Adjacent values still show edge, so it's not a single bar spike, but does periodicity as a leverage point pass your smell test, or does it look like curve fit?
3. With about a year and a half of IQFeed one-minute data per symbol, is paper trading for 3 to 6 months the right next step, or is there a sharper validation move I'm skipping?
Planning to paper trade as the next step.
Dave:
Good description and questions here.
The first thing I think about with a strategy like this is: what could make me more confident in it?
Here's what sticks out as easy things to move your confidence ball forward.
curated universe of 564 US-listed symbols
I worry when I hear you've meticulously created your symbol list for the strategy.
The fact that you've created it at this particular moment in time is somewhat arbitrary.
Create a more generic set of rules for your universe that can be applied at any given time across all US equities.
Your in-sample and out-of-sample approach doesn't seem like true IS/OOS testing. It seems like you've just split your data to see how the results changed over time.
A real IS/OOS approach would use the in-sample period, run your entire optimization process, and then test the results on the out-of-sample period.
I wouldn't obsess about the 10, 12, or 15-minute periodicity decision.
Just pick one and go with it.
A year and a half of data for the backtest might seem like a lot, but you've already done the foundational work here.
At this point, it should be trivial for you to run a backtest across a much longer period.
Here's the thing:
Would you be more confident knowing how the strategy performed across several years or just 1.5 years?
I think you know the answer.
One final thing: it's hard for me to believe that the trailing stop is improving this strategy.
It's almost certainly getting you out of profitable trades sooner than you should.
Test without it and see.
Good questions, Brad, and thanks for sharing with the group.
-Dave
P.S. Wish you could backtest your ideas - without spending months learning to code? Run your first backtest in an afternoon using my Amibroker AFL Course.