Community › Automated Trading

Automating crypto is a different discipline and I underestimated it

CryptoKarl· May 2026 ago· 👍 11· 💬 15
Stocks have a bell. Crypto doesn't, and that changes what automation is for. With stocks I automate to execute better during a window I'm awake for. With crypto I'm automating the hours I'm asleep, which means every failure mode I could babysit through is now unattended: a connection drop, a weird wick, a funding flip, a thin weekend book. My rule now is that anything running while I sleep has a hard position cap and a hard time exit, no exceptions. The strategies I'd call clever all run only when I'm at the desk.
CryptoKarl
30 posts · 0 followers
+ Follow

Related reading

Automated TradingAutomating a slow strategy feels wrong and works fineAutomated TradingAutomation doesn't fix a strategy, it just applies it fasterBrokers & APIsgateway staying up overnight — how do you nightowls handle it?Brokers & APIsnight trader here, so does the local vs cloud thing even apply to me?

15 replies

N
NightOwl_Yuki· May 2026 ago
The weekend book is the one people miss. Same strategy, same size, completely different fill quality Saturday morning.
R
RiskFirstRita· May 2026 ago
"Clever strategies only when I'm at the desk" is a rule I'd like to see more people adopt.
P
PremarketPete· May 2026 ago
- hard cap: yes - time exit: yes - and a heartbeat check so you know the bot is actually alive when you wake up Dead bot with an open position is the quiet one.
BacktestBetty· May 2026 ago
Two things I'd add from testing overnight rules: 1. Backtest your time exit separately from your entry. If the exit only looks good because of one lucky stretch, you'll find out. 2. Careful about tuning the sleep window itself. "Only trade 02:00-06:00" is exactly the kind of parameter that fits the sample and nothing else. I've fooled myself with that one.
SlowSwing_Sam· Jul 2026 ago
Or, hear me out, don't run anything while you sleep and get a proper night's rest. My positions are measured in weeks so a wick at 3am is just noise I never see. You day trading folks invented a way to be stressed 24/7 and then automated it.
MLcurious· Jul 2026 ago· edited Aug 2026 ago
Honest question because I'm still fuzzy on this. When you say hard time exit, is that a fixed clock time or a max holding duration from entry? I've been building a small model and I keep assuming the market is continuous so I never even thought about the weekend book being different. Kind of a lot to unlearn.
C
CoffeeAndCharts· Jul 2026 ago
The bell thing really is underrated, my whole routine is built around a start and an end and crypto just refuses to give me that. Good on you for keeping the clever stuff at the desk, that's the part most of us learn the hard way.
S
SpreadWatcher· Jul 2026 ago
Thin book plus a wick is not one problem, it's the same problem twice. Your stop gets filled at whatever the book offers at that second, and on a quiet Saturday that can be a long way from where you thought your risk was. Cap the size and assume the fill will be worse than your test says.
CryptoKarl· Jul 2026 ago· edited Aug 2026 ago
Max holding duration from entry, MLcurious, a fixed clock time meant I'd sometimes hold something for 20 minutes and sometimes for 9 hours under the same "rule", which isn't a rule, it's a mood. And Pete's heartbeat point is the one I added last, after the classic wake-up-to-a-position-I-don't-remember-opening experience. Sam, you're not wrong, I just have a leverage problem and a sleep schedule to match.
I
IndicatorSkeptic· Jul 2026 ago
"The market is continuous" is the assumption doing all the damage in this thread. It trades continuously, it isn't liquid continuously. Those are extremely different things and only one of them shows up on your chart.
H
HalfKelly· Jul 2026 ago
Worth adding the maths angle to the hard cap: your sizing formula almost certainly assumes you can exit near where you intend. Once fills get worse than the stop, your effective edge shrinks and your effective loss grows, so the size that was fine at the desk is oversized at 3am. I just size the overnight book at a fraction of the daytime one — half is arbitrary but it's the right direction, and being roughly right on size beats being precise about the wrong number.
L
LurkerLee· Aug 2026 ago
Do you cap total exposure across all sleeping bots, or per bot? Three small caps can add up to one big position in the same direction.
DataDrivenDee· Aug 2026 ago
Genuinely curious how you'd validate the heartbeat piece — like, do you test that the alert fires, or just trust it's wired up? Seems like the kind of safety net nobody finds out is broken until the night it matters.
GrandpaGrizzly· Aug 2026 ago· edited Aug 2026 ago
Watched a fella in the old days leave a stop order with a broker over a long weekend and come back to a fill he described, politely, as educational. Different century, same lesson: the market doesn't owe you a counterparty at the price you had in mind. Hard caps are the cheapest insurance there is.
ZenTrader_Ana· Aug 2026 ago
The line that stuck with me is "that isn't a rule, it's a mood" — that's really a feelings problem dressed up as a settings problem. Curious how you sleep now with the caps in place, has the hard exit actually quieted the urge to check your phone at 3am, or did the checking habit outlive the fix?
Sign in to reply →
← All automated trading