In crypto, “copytrading” means automatically mirroring another trader’s positions. They buy, you buy; they sell, you sell. You outsource the decision to someone whose judgement you trust more than your own.

This is an underrated life strategy.

A friend recently told me he wished he had started using Claude Code when I first mentioned it two months earlier. I had sent him a video calling it “pretty amazing” and saying “this is gonna be crazy.” He did nothing, and two months later he was kicking himself.

The right move was to try it immediately. Trying something recommended by a smart person usually costs little, whereas ignoring it can cost months.

Why smart people copy harder

You might expect my dumbest friends to copytrade me most because they need the help. I see the opposite: my smartest friends copytrade me hardest.

The impolite explanation is that dumb people are not smart enough to work out whom to copy. They cannot reliably distinguish good judgement from confident bullshit, track predictions, or notice who is consistently right. Pride and insecurity can entrench the mistake later, but the first problem is pattern recognition: they have not identified anyone worth copying.

Smart people have spent years calibrating. They know whom to trust on which subjects, so when someone in that network makes a high-conviction recommendation, they act.

The word just in “life is at least 50% just figuring out who to copytrade” is doing ironic work. Choosing well might be the hardest epistemic skill there is. You have to follow recommendations over time, separate judgement from salesmanship, recognise that competence is domain-specific, and update when a track record changes. Then you have to ask whether you are the person with poor judgement about good judgement.

That last part makes the problem recursive. Confidence in your choice of people rests on confidence in your ability to evaluate them, which is another judgement call. The problem is that 90% of people have shitty judgement about who is worth copytrading, and they do not know it.

Some useful heuristics

Follower counts are mostly a distraction. Patrick McKenzie has 423 YouTube subscribers and is one of the highest-signal people in tech and finance. Filtering by popularity selects for entertainment value, not judgement quality.

Competence is also local. Someone who gives excellent technical recommendations might be an idiot about relationships; copy the demonstrated competence, not the person wholesale.

If you are not cutting back at least 10%, you are probably not copytrading aggressively enough. That is adapted from Elon’s thing about firing, and I have never seen someone copytrade the right people too much. If your 20s were spent working out who deserves attention, your 30s are the time to use that knowledge rather than endlessly shop for new sources.

One friend recently told me he was narrowing his information diet to Zvi, ACX and a few others. He had done the work of deciding who was worth reading and wanted to harvest that investment.

There is also a two-hour test. If you have spent two hours with someone and still have no view on whether they are copytrade-worthy, you are neglecting a basic life task. You should be evaluating all the time. If you remain in stasis, they will eventually cut you off, and should, because you are wasting their time.

The asymmetry favours trying

The downside of trying a smart person’s recommendation is usually capped: install the software, read the book, or use the technique for a week. The cost of ignoring it may be months of productivity or a career insight you never recover. The upside stays open-ended.

I do not think this becomes actionable merely because I have explained it. Telling someone with poor epistemics to track predictions may leave them tracking the wrong things. “Beware follower counts” can become permission to follow a fringe crank, and one lucky call can be enough for them to double down on a bad health influencer.

The heuristics only work if the underlying skill is already there, which makes this less advice than a description of what smart people do naturally. The most I can hope is that smart readers recognise the pattern and lean harder on the people they have already identified. Everyone else will continue to copytrade poorly, if at all. I think that bleak conclusion is true.

If it makes you uncomfortable, though, the discomfort may create a useful moment of doubt:

  • How would I know if I am wrong? If you cannot say what would change your mind about someone, you are believing rather than tracking.
  • What do my smartest friends think of this person? Ask the friends who push back, not the agreeable ones. If they think your guru is a clown, that is data.
  • Would I bet money on the next prediction? Use real money, enough to hurt. If not, why are you following the advice?
  • When did this person last admit an error? People with good judgement update; people performing confidence do not.

I cannot give you good epistemics, but those questions may put doubt in the right places.