Community › Strategies & Backtesting

adapting those four exit variants to 24/7 crypto?

NNightOwl_Yuki· Jun 2026 ago· 👍 4· 💬 6
just finished reading the piece comparing four intraday exits against the same entry, the one with the fixed target, the time exit, the give-back rule and the breakeven bump. really like the framing of holding the entry constant, i've honestly never done that properly, i just tweak everything at once and then wonder why i can't tell what helped lol. but it's written for US stocks and i trade coins at like 3am so a couple of things i'm unsure about. 1. the time exits (1200s, 1800s). in stocks a hold time kind of borrows meaning from the session, there's an open and a close and dead midday. crypto just... keeps going. so is a time exit still doing the same job for me, or is it just a random timer? my instinct is it still cuts the trades that went nowhere, which is most of my bad ones tbh, but i don't know if i should scale the number to volatility instead of clock time. 2. the -1.0 stop. on the pairs i trade that's noise, i'd get stopped constantly. if i widen it to say -2 or -2.5 do i have to widen the target and the give-back threshold by the same factor to keep the comparison honest? or does scaling everything just turn it back into the same test. 3. variant 3 vs 4 is the one i actually care about. give-back protection lets a winner breathe a bit but pays for it, breakeven-after-1.5 basically says never let a winner turn red. at night my moves tend to be one clean push then a long chop, which feels like it favours the give-back rule, but i genuinely don't know, i'm guessing from vibes. anyone run these side by side on crypto? mostly wondering how many trades i need before the difference between 3 and 4 means anything and isn't just me reading tea leaves.
N
NightOwl_Yuki
27 posts · 0 followers
+ Follow

Related reading

Strategies & Backtestingdo those four exit variants translate to crypto overnight?Strategy LibraryDay trading exit rules: four variants comparedAutomated Tradingthe four exit variants aren't quite apples to apples, right?Strategies & Backtestingconfused about how to actually compare the four exit variants

6 replies

F
FiveMinFiona· Jul 2026 ago
On (2): if you scale the stop, target and give-back all by the same factor you are basically just re-expressing everything in R units, and yes, it's close to the same test — which is fine, that's the point of it. What actually changes the answer is scaling them by *different* amounts, e.g. wider stop but same target. So pick your stop first from what your pairs actually do, then treat the rest as multiples of it and keep those multiples constant across the four variants. On (1), I'd keep clock time as the baseline and add a volatility-scaled version as a fifth variant rather than replacing it. Then you can see whether the scaling earned anything.
CryptoKarl· Jul 2026 ago
3am alt coin chop is its own beast lol. one thing nobody warned me about: on perps a time exit is not free, you're paying funding on the hold, so a 1800s exit and a 300s exit are not the same trade even if the price ends identically. matters more the more leverage you're carrying, ask me how i know
Q
QuietVol· Jul 2026 ago· edited Aug 2026 ago
The reason holding the entry constant matters so much here is that variants 3 and 4 are then evaluated on the *same* trades, so you can compare them pairwise. Look at the per-trade difference (give-back result minus breakeven result) rather than two separate averages. That difference is zero on every trade where price never reached +1.5 and never gave back enough to trigger either, so your effective sample is much smaller than your trade count. On my equity stuff maybe a third of trades ever separate the two rules. Have you checked what fraction of your night trades actually reach the 1.5 trigger at all?
N
NightOwl_Yuki· Jul 2026 ago· edited Aug 2026 ago
yeah the funding thing hadn't even crossed my mind, that alone probably kills the 1800s variant on some of these. and Fiona ok that makes sense, so scaling everything uniformly isn't wrong it's just not new information, got it. I'll fix the stop from actual noise and express the rest as multiples. QuietVol that reframing is really useful, i was about to count all my trades as sample. no idea what fraction touch +1.5, will go look, i suspect it's low which is kind of the whole answer isn't it
DataDrivenDee· Jul 2026 ago
The bit I'd poke at is "one clean push then a long chop" — how would you actually check that? Because if it's true you should see it in the excursion data before you ever run the exit variants: record max favourable excursion and where in the hold it happened, per trade. If the peak clusters early and then price drifts back, that's the give-back rule's case made without needing to simulate variant 3 at all. If peaks are scattered through the hold, it was vibes. Cheap thing to log either way.
SlowSwing_Sam· Jul 2026 ago
the good news is at 3am nobody can tell you what the session structure is supposed to be, so you get to define it yourself. downside is you also have to define it yourself for what it's worth I hold for days and the give-back vs breakeven question doesn't go away, it just runs slower. sleep is undefeated though
Sign in to reply →
← All strategies & backtesting