the time window skeleton — how do you pick the times without fooling yourself

So I read the time-windowed skeleton thing in the strategy library over the weekend, the one that's basically "wrap your idea in two clock times and force yourself flat before the close." And my first reaction was mild embarrassment, because the forced flat line is exactly the thing I have never had, and I have the account history to prove it. My best example is a mean reversion idea I ran for about three months that was fine, genuinely fine, right up until it wasn't — I never held overnight *on purpose*, I just didn't have a rule that said get out, so a couple of times a month it held by accident and those accidents ate everything the good days made. So that part I got, loud and clear, and I've already gone back and bolted an exit-by-clock onto the two things I still run. The part I'm less sure about is picking the window in the first place. The write-up says the open is fast and expensive, midday is slow, the last half hour is its own animal — fine, I believe that, it matches what I've felt. But the moment I sat down to actually choose numbers I realised I was going to do the dumb thing: run it at 09:45, run it at 10:00, run it at 10:15, and pick whichever start time made the equity curve prettiest. That's not a window, that's just a parameter I've given a respectable-sounding name. I've done this before with moving average lengths and it did not end well. So the question for people who actually use this shape: do you decide the window from a reason (this idea needs volume, volume shows up after the first move settles, therefore X) and then leave it alone? Or do you test a handful and accept you're fitting a bit as long as the neighbouring values aren't garbage? I'd also like to know what you do about the dead zone — entries stop at say 14:30 but the flatten isn't until 15:45, so there's an hour where you can only be losing or leaking. Do you shrink that gap or is the gap the point? Also, minor thing, but the exit block puts the forced flat first and everything else in the elif. Am I right that the ordering there matters more than it looks?