Last March my bot sat out four days of a clean-looking move. Price ripped, my feed lit up, and I spent most of that week typing a message to myself that began with 'this rule is broken.'
The rule was a cooldown. After three losing trades in a row, no new entries for 24 hours. I wrote it in 2022 after a bad month, when I was calm and a little ashamed. By last March I hated it.
It's the first one I wanted to delete. That should have told me something.
Resentment is information
Think about when a rule actually bites. It bites at the moment you want to do the thing it forbids. A rule against revenge trading only costs you anything when you feel like taking revenge.
So the rules that feel worst are the ones sitting directly on top of your worst impulses. Nobody resents a rule that never interferes. The sting is the rule doing its job.
I'm not saying every annoying rule is a good one. Some are just badly designed. But irritation by itself can't tell you which kind you have, and that's the trap.
What the cooldown actually cost
When I finally went back through the logs, I looked at every time the cooldown had fired. Eleven times over about two years. In those windows, I tagged what the bot would have done had it kept trading.
Mixed result. A few of those windows contained decent entries I missed. Several contained the exact choppy, whipsaw conditions that produced the three losses to begin with. Net, the cooldown was roughly a wash on returns and clearly better on drawdown. Small sample, so take it loosely.
But look at the shape of that. Not a disaster, not a miracle. A rule that trades a little upside for a lot less pain. And I was ready to rip it out because of one vivid week I could still feel.
Why we delete the wrong rules
The problem isn't that traders are lazy about rules. It's that the cost of a rule is visible and the benefit isn't. The trade you skipped shows up as a chart you can stare at. The loss you avoided shows up as nothing.
So you remember every move you missed and none of the blowups that never happened. Memory is a terrible auditor. It only records what you can see.
There's a second problem. You review rules when you're annoyed, which is exactly when your judgment about them is worst. Deleting a rule after a frustrating week is the same impulse the rule was written to block, just wearing a more respectable coat.
A test before you kill a rule
Here's what I do now. When a rule makes me angry, I write down what it prevented, with dates, from the log. Not what it blocked in the one trade I'm upset about. All of them.
Then I wait a set period before touching it. For me that's two weeks. Change it on a Tuesday afternoon, not in the heat of Sunday night, and only if the full record says it's costing more than it saves.
If the numbers say the rule is wrong, change it. Good. That's a real edit, made with evidence. If they say it's fine and you still hate it, congratulations, you probably found the one worth keeping.
I think this is the quiet argument for automation. A bot doesn't resent anything, so it can't talk itself out of a rule at 2am. If you'd rather not be the person enforcing yours, the full rules and backtest data for our bots are at v33systematic.com.