Run 3 Hit the Kill Switch: Down −3.2% and Done for the Day
Run 3 of our public paper-trading experiment lasted about 13 hours. It lost money. And at −3.2%, it did the thing it was designed to do: it stopped itself.
Final equity: $967.97
P&L: −$32.03 (−3.2%)
Trades: 7 · Wins: 1 · Losses: 5
Fees paid: $2.71
The trades, in order
One winner carried the day: koinu doubled within the hour, and the bot's 1-hour fast exit locked it in at +$12.38 (+101%). That's the 2x scale-out doing its job — principal back, rest riding free.
Everything else bled. CATONICE exited flat at −2.0% (−$0.73). MINTER ran straight into the −50% stop and was cut at −52.6% (−$13.26). The remaining four trades were all red, and the math is blunt: the one winner made $12.38 while everything else netted −$44.41.
The guardrail fired at −3.2%
The bots run a daily loss limit. The moment the day's P&L crossed −3.2%, the log recorded one line — "Daily -3.2%: no new entries for 24h" — and all open positions were flattened. No human judgment, no "let's wait and see," no hope. The system decided it was done, and it was.
That is the entire point of paper trading in public. Anyone can write rules; what matters is whether the rules survive contact with a red day. The Safe and Medium tiers didn't even take a trade — they sat the day out. The High tier traded, lost, and got benched by its own risk limit.
What this proves (and what it doesn't)
The daily loss halt fired exactly as coded, mid-run, with no one watching. That's the behavior you want before real capital is anywhere near the system.
One winner in seven trades doesn't prove anything except that small-sample runs are noisy. Three runs in, the honest verdict is: the exits function, the entries are unproven, and real money stays off the table.
The dashboard carries all three runs — the losses, the fees, the kill-switch line — for anyone to audit. When a strategy finally earns real capital, the proof will be there in public first.
Every trade, every fee, every halt — on the public dashboard.
Open Nova Trader →