the two book-ratio lines: is that asymmetry doing real work?

Read the order flow filter piece in the strategy library over the weekend — the one that sits in front of an entry and asks whether the tape agrees before you're allowed to press the button. Nicely written, and I liked that it's honest about not being a strategy on its own. The bit I keep circling is the pair of book-size ratio conditions. The author explains they aren't redundant: one rejects a book stacked against you, the other rejects a book stacked *for* you (too lopsided the friendly way, presumably because that's often a wall about to get pulled or a queue you'll never get filled in front of). Fine, I follow the logic. What I can't work out is the asymmetry, the two allowed ratios are quite different sizes, one much looser than the other. Is that an empirical finding from whatever names were tested, or is it a structural claim about how books actually look on the bid vs ask side? Because those two answers imply very different things about whether I should copy the numbers or refit them. My own interest is downstream of the entry, as usual. If a flow filter genuinely raises hit rate, the honest response isn't to trade bigger per position — it's to notice your edge estimate now has fewer samples behind it, since you're rejecting most bars. I ran the conditions as a pure counter for a couple of weeks on a handful of liquid names, no orders, just tallying how often all four lines were true at once. It fired far less often than I'd assumed from reading the post. Which is arguably the point of a filter, but it does mean any win-rate figure you get out of it in a month is close to meaningless for sizing purposes. So two questions for anyone who's actually run this: 1. Did you keep the asymmetric ratios or make them symmetric? Did it change anything you could measure? 2. How long did you have to let it run before the sample felt big enough to size off, rather than eyeball? Also mildly curious whether the volatility floor is doing most of the rejecting and the book lines are along for the ride. Would be easy enough to test by turning them off one at a time, and I suspect someone here already has.