the angle scaling bit in that crypto momentum writeup - questions

Read the crypto momentum entry piece in the strategy library a couple of times now and I want to poke at it before I put any of it near a live account. Numbered, as usual, sorry. 1. The ChangeAngle scaling note is the most useful paragraph on this whole board and I say that as someone who has wasted actual weeks not understanding it. I had been treating the angle like a normalised thing, i.e. "20 degrees is 20 degrees". It isn't. Widening the window while keeping the threshold means you have silently written a condition that basically never fires, and then you sit there wondering why your backtest has 4 trades in it and one of them looks amazing. 2. So my question: is there a sane way to pick the pair? What I've been doing is fixing the window first based on how long I actually want the move to have been building, then sweeping the angle until trade count lands in a range I can defend statistically. But that is dangerously close to fitting the threshold to the sample, which is the exact thing I nag everyone else about. 3. On the hold cap. The piece is blunt that the time exit is what ends a bad trade when nothing else will, and I agree with the logic completely. What I don't have a feel for is whether 3600 is a real number or a round number. In a swing-ish momentum context an hour is either "plenty" or "you just cut the winner off at the knees" and I honestly can't tell which from the writeup alone. Has anyone varied it? And more importantly, did the result survive when you moved to a different chunk of history? 4. Liquidity floor. Filtering on daily amount makes obvious sense but it's also the parameter most likely to be quietly doing the work of "only trade the big names". If I lower the floor a lot, do I get more trades or do I just get more trades I'd never want filled? 5. Standing warning to myself and anyone else: three parameters, each with a plausible range, is already enough surface area to fool yourself badly. Two of these interact (window and angle) so it's really a grid, not three sliders. I'm holding out a block of history I have not looked at yet before I touch anything. Anyone who has actually run something shaped like this - which of the four knobs turned out to matter and which were noise?