01The problem
The biggest risk for someone investing alone is not a wrong analysis, but a decision made on impulse. A system is only worth something if it applies the same rules even when you would be tempted not to.
02How it is used
- The rules are fixed beforehand, in writing: when to enter, when to exit.
- The agent checks the market at regular intervals, day and night.
- It buys only with three gates open together. Otherwise it writes the reason in the log and waits for the next check.
- You read the log. Every decision, even the decision to do nothing, has its own line.
03How it is built
- Gatesthree independent conditions: the global risk climate, how crowded the bet is, how fresh the signal is
- Executionsimulated (paper trading), no real capital
- Exita trailing stop decided in advance
- Modelreads the news and writes a judgement in the log, with no power over the decisions
- Verificationa second model challenged the code: 32 findings, 18 fixed
04Results
In the first 44 days there was no purchase: the three gates were never open together. The signal fired twice and was rejected both times, because the context wasn't right. With only one closed trade, the return says nothing, so I don't publish it.