The best decision maker is a model of you

Two ways to be wrong

I read Daniel Kahneman’s Noise recently (with Olivier Sibony and Cass Sunstein) and one distinction has stuck with me since.

Error isn’t one thing. It splits into bias and noise, and they aren’t the same problem. Bias is the familiar one: your reasoning points the same wrong direction every time. In trading that’s FOMO or panic selling, pushing you toward buying high and selling low. There’s already a post on this blog about that half of it.

Noise is quieter. It isn’t a direction, it’s the absence of one: two people, or one person on two different days, facing an equivalent case and landing somewhere different, for no defensible reason. The book’s core examples aren’t about trading at all. Insurance underwriters pricing the same policy come back with premiums that differ by far more than anyone at the company would guess. Judges hand down different sentences for comparable cases depending on things as arbitrary as the weather or how their morning went. Doctors shown the same scan twice, months apart, without recognizing the repeat, sometimes disagree with their own earlier read. None of them are behaving badly. They’re just far less consistent than they think they are.

Trading is the version of this I actually get to do something about, so that’s where the rest of this goes. But the pattern is bigger than crypto and once you’ve seen it in underwriting and sentencing, it’s hard not to notice it in your own decisions too.

A model of you beats you

One chapter makes a claim I had to reread several times: a simple statistical model built from your own past decisions usually outperforms you making the same decision live. Not a smarter model or more information than you already had. The same judgment you already carry, just applied the same way every time.

The model doesn’t win because it’s wise, but because it’s boring. It doesn’t have a bad night. It doesn’t get more confident after two good calls in a row. Every time the inputs match, it produces the same output and that consistency alone turns out to be worth more than most people expect.

That’s close to what a written trading rule is. “Buy when price drops 5% in a session, sell at 8% recovery, never more than €10 a position” isn’t a smarter version of your judgment. It’s your judgment from your calmest hour, applied without drift. You’re not being replaced by something better than you. You’re being protected from your own “noisier” moments.

Simple beats clever

The next chapter goes further. Robyn Dawes’ research on “equal-weight” models, plain formulas that weigh a handful of factors equally with no fine-tuning, kept beating specialists hand-weighing the same factors with years of experience behind them: parole boards predicting reoffending, universities predicting student performance, clinicians predicting how a patient’s condition would develop. Not by a little. Reliably.

The specialists weren’t careless. They were doing what feels like the responsible thing: weighing context, adjusting for nuance, treating each case as somewhat unique. Kahneman calls the trap this creates the “illusion of validity.” The more factors you juggle, the more in control you feel, and the worse you tend to do, because you’ve made yourself inconsistent in a far more complicated way.

I recognize this one directly, from my own experience testing plenty of price indicators. Building a strategy in Aurono, it’s tempting to keep adding conditions: one more indicator to catch the edge case, one more filter for “unless the market is doing this.” Each one feels like it’s making the rule smarter. Usually it’s making the rule harder to apply the same way twice, which was the one property doing all the work.

What rules are actually for

None of this makes anyone a better judge of anything, and I don’t think that’s the book’s argument. What a rule does, in a courtroom, an underwriting desk, or a trading strategy, is take the judgment you already have, at your calmest, and stop it from drifting. Not because Tuesday-you is wrong and Thursday-you is right, but because the two of them disagreeing on an identical case was never doing anyone any favors.

That’s most of what “buy rule, sell rule, position size” actually buys you: not intelligence but consistency. Kahneman’s book is the first place I’ve seen that argument made with real research behind it, instead of as a trading platitude. And it helps, of course, that Kahneman won a Nobel Prize in Economics…

If you’ve never written your rules down, this is a fine place to start: three lines, no fine print. The day you’d have made a different call on the same setup twice is the day you find out how much noise was costing you.

→ See how shadow mode tests your rules for free · Read the case for rule-based trading