Community › Automated Trading

widening the angle window without killing the entry - sanity check

NNightOwl_Yuki· May 2026 ago· 👍 3· 💬 5
hey all, night shift Yuki here ☕ read the crypto momentum writeup in the strategy library the other night (the one with the liquidity floor + hold cap) and it explained something that had been bugging me for weeks. i had a version of that buy condition running with the window pushed out to 600 because 60 seconds felt too twitchy for me, and i kept the angle at 20 because... i mean why would i change it, right. it basically never fired. i assumed my filters were just picky. turns out i was quietly asking for a move like 10x bigger than i thought i was. so now i get *why* it stopped triggering. what i don't have a feel for is where to land. the piece gives the two reference points (30 degrees over 60s vs over 600s) which is enough to see the shape of it, but not enough for me to guess what a sane 300 second setting is. do people just scale it down until the fire rate looks normal over a week of paper, or is there an actual number folks converge on for the 5min-ish window? i don't want to overshoot and end up buying every little wiggle either, that's the opposite problem. other bit i'm chewing on: the hold cap. mine runs while i'm asleep so i take the point completely, nothing else ends a bad trade. but 3600 on a swing-ish idea feels short? like the whole reason i'm on the longer window is that i want the move to have some room to breathe. is the right move to stretch the cap and tighten the stop instead, or does that just mean i'm holding losers longer with extra steps. also curious if anyone runs the liquidity floor higher than the example. i've been sticking to bigger pairs anyway so it rarely binds for me, wondering if that means i should raise it until it actually does something.
N
NightOwl_Yuki
27 posts · 0 followers
+ Follow

Related reading

Automated Tradingthe angle scaling thing in that crypto momentum writeupStrategies & Backtestingthe angle scaling bit in that crypto momentum writeup - questionsAutomated Tradingthe angle window math from the sleep-safe crypto writeup — am i reading it right?Strategies & Backtestingthe angle/window scaling thing from that crypto momentum writeup

5 replies

CryptoKarl· Jun 2026 ago
the fire-rate-until-it-looks-normal approach is basically what everyone does whether they admit it or not, but do it on the same pairs you'll actually trade because the number that feels fine on majors will spam you on thinner stuff. re the hold cap - stretching it while tightening the stop is a real tradeoff not a free lunch, you're just moving where the trade dies, and personally i'd rather it die on time than on a wick at 3am. and yeah if your floor never binds it's not a filter, it's decoration.
H
HalfKelly· Jun 2026 ago
The scaling is easier than it looks if you stop thinking in degrees for a second. Whatever the engine does internally, the angle comes from a slope, so tan(theta) is proportional to (move / window). Same move over 5x the time = one fifth the slope. So the 60s equivalent of your 20 degrees at 300s is arctan(tan(20)/5), which is about 4 degrees. At 600s it'd be a hair over 2. That's why your 600/20 combo basically never fired - you were asking for roughly ten times the move you had in your head. I'd treat that 4 as the anchor, not the answer. Run it slightly above and slightly below and see what the fire rate does, because the arctan curve is nearly linear down there, small changes in degrees are big changes in required move. On the hold cap: if you stretch it you're increasing time-in-market per trade, which means fewer independent bets per week for the same capital. That has to come out of size somewhere, otherwise you've quietly levered up on the same account.
C
CoffeeAndCharts· Jun 2026 ago
oh this thread is going straight in my morning read pile, the arctan thing is the cleanest explanation of that setting i've seen anywhere. welcome to the angle rabbit hole Yuki ☕
N
NightOwl_Yuki· Jul 2026 ago
ok the arctan bit just made something click, thank you. i'd been treating degrees like they were a difficulty dial instead of a slope and that's exactly the mistake. and yeah Karl's point about testing on the pairs i actually trade is fair, i was lazily eyeballing fire rate on whatever was in front of me. Karl your hold cap line got cut off btw, curious what the rest of it was - stretching while tightening the stop was exactly the thing i couldn't reason my way through.
Q
QuietVol· Jul 2026 ago
One small reframing on the fire rate: it isn't a number, it's a distribution, and a week of paper gives you very few draws from it. If your setting fires say a handful of times over that week, the variance on that count is large enough that a slightly different week would have you tuning to a different answer entirely. I'd want to see the spread across several non-overlapping weeks before calling any threshold 'normal'. On the cap - the useful thing to look at isn't whether 3600 feels short, it's the distribution of time-to-favourable-excursion for the entries you did get. If most of the good ones resolve well inside the cap, stretching it only adds tail exposure. Do you have the per-trade timing data from your paper runs, or just the outcomes?
Sign in to reply →
← All automated trading