how do you port the 'trades less than you expect' filter to 24/7 crypto?

hey all, been chewing on that selective momentum entry in the strategy library (the US stocks one with the deliberately restrictive buy block) since last night's session, which for me is everyone else's bedtime. the part I actually agree with hardest is the framing at the top. Most losses come from trades you took because nothing was happening. that is literally my whole problem. I sit there at 3am, chart is flat, I get bored, I take something marginal, and that's the trade that ruins the night. so the idea of a condition set that just sits still and doesn't fire is very appealing to me. but two chunks of it don't map onto crypto and I'm not sure what the honest replacement is: 1. the time window (10:15 to 14:15). that's clearly doing real work — skipping the open, skipping the last hour. crypto has no open. I *could* just carve out some hours, but that'd be me inventing structure that isn't there. or is it there? like, is there a defensible "don't trade this stretch" for crypto or is that just me pretending my timezone is meaningful? 2. DayAmount > 20000000. in stocks that's a liquidity floor and it resets daily. on a 24/7 asset a rolling day figure never really resets, it just drifts, so the same number means something different at 3am than at 3pm. do you guys use a rolling window volume instead, or a relative measure (volume vs its own recent average) rather than an absolute? and the angle bit, the note about degrees scaling with the window was genuinely useful to me, I'd been treating ChangeAngle as if the number meant the same thing regardless of n. but overnight crypto moves are noisier and thinner, so I suspect a 60s angle threshold that's selective in a stock session would fire constantly on a low-liquidity book, on moves that go nowhere. instinct says raise it, but then I'm back to the exact loosening/tightening fiddling the author warns about, just in the other direction. hasn't been an issue in testing yet because honestly it's barely fired. which I'm told is the point. anyone actually run something this restrictive overnight and stick with it?
N
NightOwl_Yuki
27 posts · 0 followers
+ Follow

Related reading

7 replies

CryptoKarl· Jul 2026 ago· edited Aug 2026 ago
the boredom trade is the whole ballgame, I don't even need the rest of the post. on the volume thing I'd go relative every time — some ratio against a trailing average, because an absolute floor on a 24/7 book just means "trade during US hours" with extra steps, and you already said you don't want to pretend your timezone is special. and yeah don't raise the angle to compensate for thin books, that's just you paying to be told the same noise twice, either accept it fires rarely or accept it fires on junk.
MLcurious· Jul 2026 ago
so if you go relative, is the trailing average itself a problem? like if the whole book has been quiet for 6 hours the average drops with it and suddenly quiet looks normal again. or is that fine because you also have the angle condition doing work? genuinely asking, I keep tripping over this in my own stuff
GrandpaGrizzly· Jul 2026 ago
Son, the crypto crowd rediscovers this every cycle. There IS structure in the 24/7 tape — it's just borrowed. Asia handoff, London, the US cash open, futures settle days, weekends when the desks go home. It isn't a bell but it's not nothing either. Where people go wrong is carving hours out because they feel tired, not because the book thins. Log which hours actually gave you the bad fills for a couple months and let the data pick your dead zone. Stay humble.
H
HalfKelly· Jul 2026 ago
The line I'd hold onto is "it's barely fired" — that's not a bug report, that's the filter doing its job. Worth remembering the arithmetic: if a rule takes 4 trades a month instead of 40, you need roughly ten times as long to say anything honest about its edge. So budget for a long, boring evaluation window rather than loosening it because the sample feels thin. Loosening to get sample size is how you end up measuring the boredom trades again.
N
NightOwl_Yuki· Aug 2026 ago
yeah the trailing-average-drifts-down thing is exactly what's been bugging me MLcurious, no clean answer yet. and Grandpa that's a fair reframe, borrowed structure. I do notice fills get ugly in a specific stretch, I just never wrote it down, only remembered it. logging it is the obvious move and I've been not doing it for months lol
R
RiskFirstRita· Aug 2026 ago· edited Aug 2026 ago
Before you touch a single threshold: write down what the rule is supposed to skip, then check whether the trades it skipped would actually have lost. Otherwise you're tuning on vibes at 3am, which is the worst possible time to tune anything. And whatever number you land on, size the 3am ones smaller than the ones you'd take awake, thin book, tired operator, same position size is not the same risk.
ZenTrader_Ana· Aug 2026 ago
What strikes me is that the filter is already working on the part you said was the real problem — you sit there and nothing fires. So the discomfort isn't with the rule, it's with sitting. How does that hour feel when the chart is flat and the block stays silent? For me that's the actual skill, and it doesn't live in the parameters.
Sign in to reply →
← All strategies & backtesting