Hi all,

We run an intraday opening range breakout on liquid US stocks through IBKR (paper account for now). The backtest is fine. The live paper fills are not, and almost all of the gap comes from the entry.

The setup in general terms: when price breaks the opening range, we enter in the breakout direction with a protective stop on the other side of a fairly tight band. We measure everything in R, where 1R is the planned risk per share (distance from the breakout level to the stop). The stop is tight, so 1R is often only a few cents, and a cent or two of spread is already a big piece of it.

What we measured:

Entry slippage vs the breakout level. Over three weeks of paper trading in September and October the median entry was 0.47R worse than the level, and the mean was about 0.87R (a few bad fills on fast names pull it up). Our own math says the strategy breaks even at an average entry cost of roughly 0.23R. So we pay about two to four times what the edge can afford.

Faster sending. We cut the time between the signal and the order reaching IBKR. In a tick replay of 344 paper entries this was worth about $4 per trade. It helps, but in the first live paper session with it the median was still 0.62R (16 fills, one day, so treat that as noise for now).

Stop trigger methods. We replayed IBKR's Double Last, Bid/Ask and Midpoint trigger methods against plain Last. All three came out slightly worse than Last on our data.

Limit at the midpoint. Instead of a marketable order we post a limit at the mid, capped at 0.5R past the level, wait 30 seconds, and either fall back to a market order or skip the trade. When it fills it saves roughly $3 to $14 per trade. But 20% to 55% of the time it does not fill, and the ones that run away without us are often the strongest breakouts. In the tick replay the net result changed sign depending on how we model fills. On the first out-of-sample day it was between minus $0.40 and plus $2 per trade with market fallback, so we cannot tell it from zero yet.

A spread filter (skip names where the spread is too wide relative to R) looks good or bad depending on the threshold, and we picked the threshold on the same trades, so we do not trust it. It also changes the strategy rather than the execution.

The part we cannot test well: the IBKR paper account does not simulate hidden liquidity at the midpoint. Paper fills a mid limit only if the quote or a print actually crosses it, so it is probably too pessimistic. Our backtest assumes fills at the level, which is too optimistic. The truth is somewhere in between and we have no clean way to find it without real money.

Questions:

  • Which order types have actually lowered breakout entry cost for you on US equities? We are looking at IBKR Adaptive (Urgent vs Normal), IBKR Midprice, pegged orders (peg to mid, peg to primary), and routing to IEX for the midpoint peg. Any real numbers would help a lot.
  • Stop limit instead of stop market: what offset worked, in ticks or as a fraction of the spread or of ATR? How did you handle the trades where the limit was skipped?
  • Does splitting the entry help (part at the break, part on a retest or after a few seconds), or does it just turn into a different strategy?
  • Is entering a few seconds earlier or later than the first minutes of the session worth it, given how wide spreads are right after the open?
  • How do you test entry cost honestly without live money? Tick replay with a queue position model, a small live size just to measure fills, something else? How did your paper or simulated fills compare with real fills later?
  • What are we missing? Maybe the answer is that a stop this tight is simply not tradable at retail cost, and the fix is in the strategy, not the order.

We are happy to share more of the measurement method if it helps. Thanks for reading.