Actionable insights — Avoid These 10 Mistakes

1. Optimise for fewer errors, not more insight

1. Before hunting a new edge, audit the last several decisions that cost money and classify each as an error of process rather than of luck.
2. Accept the implication: the highest-return activity is usually removing a recurring mistake, not adding a new source of alpha.
3. Build the checklist from named biases so an error can be diagnosed rather than merely regretted.

• Errors that are invisible because they never became positions — the opportunities passed over for the wrong reason leave no trace in the P&L.

2. Run the named-bias checklist against your own current thesis

1. Take the ten biases as prompts and ask each as a question about a live holding: am I in denial about bad news? Am I defending this because I said it publicly (consistency bias)? Am I holding it because others hold it (social proof)?
2. Add the source-side ones: whose incentives produced the information I am relying on; am I deferring to an authority; am I believing someone because I like them?
3. Add the reaction-side ones: am I over-reacting to a small loss (deprival super-reaction); am I trading for the hit rather than the return (gambling)?
4. Finish with the method check — man with a hammer: am I applying my one favourite framework to a problem it does not fit?

• Using the checklist to explain away disconfirming evidence ("the market is just anchoring") — the biases apply to you first, the crowd second.

3. Treat "ask someone who knows" as a debiasing tool, not a shortcut

1. Identify the part of the thesis that sits outside your circle of competence and name it explicitly.
2. Find someone with operating experience of that part and ask them the narrow question, not the investment question.
3. Weigh the answer for incentive: an expert with a position is a source, not an oracle — which is the incentive-bias item on the same list.

• Authority bias arriving through the back door — the expert's confidence is not evidence, only their reasoning is.

4. When a name hits its cheapest multiple ever, name the single fear doing the work

1. Establish the all-time-low multiple as a fact, not an impression — the forward PE against its own history.
2. Write down the one thing the market is pricing. If you cannot state it in a sentence, you have not found it.
3. State your conclusion as a conditional on that fear, so the thesis is falsifiable: "if it doesn't [happen], it could be a great buy at current prices."
4. Check whether the company is retiring shares at the low — a buyback into a mistaken derating multiplies the correction.

• A cheapest-ever multiple that keeps getting cheaper — the conditional needs a review date, or it becomes the consistency bias the same issue warns about.