The Day the System Did Nothing Was the Most Important Day
April 11. I turned the system on for the first time.
10 AM. The buy scan ran. Nothing happened. KOSPI was below MA20. The regime filter blocked every buy across the board. The system just waited.
Honestly — in that moment, I thought something was wrong. I described the situation to the AI and asked it to check. It wasn’t a bug. It was the intended behavior.
That was the most important moment of the entire week.
Not How to Buy — How Not to Buy
When I first designed the system, I focused on how well it buys. Entry conditions, RSI, deviation from MA20. Most of the time went into the buy logic. But when I actually ran it, the key was somewhere else entirely. It was about when not to buy.
Not holding on through a downturn. Not entering in the first place. The regime filter is the mechanism for that. The discipline of not touching the keyboard even when you’re watching the screen — the system does that for you.
My Role Was to Document What Happened
There were bugs too. The system misread buy amounts as losses. Cash balance got stuck. I wrote down what I observed and handed it to the AI. The AI identified the cause. I verified the fix. That was my role — nothing more.
Today, the Trailing Stop Fired for the First Time
Kakao dropped ₩1,050 from its peak. Auto-sell. +3.01%. I didn’t check whether Kakao kept climbing afterward. I know that checking would shake me.
Monday, Live Account Starts
There’s still something I don’t know. When a stop-loss fires in live trading for the first time — whether I’ll want to intervene. Whether I can hold back. That one I’ll have to find out by doing it.
If the person who built the system can’t trust the system, the system doesn’t exist.
What to watch next: How I react when the stop-loss fires in live trading for the first time.
Reader question: Has there been a moment when you found it hard to trust your automated system’s judgment?