The pre-mortem

There is a strange thing about investing losses: the reasons are almost always visible before the money is committed, and almost never admitted until after it is lost. Six months on, the reader who is nursing a fall can list the warning signs fluently — the receivables were already stretching, the promoter was already pledging shares, the plant was always going to be late. The honesty is real. It has simply arrived too late to be worth anything.

The pre-mortem, an idea from the psychologist Gary Klein, is a small trick for moving that honesty to the front. It works like this. Before you buy, you sit for a moment and imagine it is a year from now and the position has failed — not "might fail", but has failed, badly, and you are looking at the wreck. Then you write down every reason it failed. Not the risks you can bear to think about; the ones the failure forces you to name. The exercise feels slightly absurd the first time, because you have to pretend the thing you are excited about has already gone wrong. That small act of pretending is exactly where its power lives.

This module is about turning that imagined failure into something you can actually use. Because a list of fears, however vivid, is not yet a defence. The pre-mortem only earns its place when each cause becomes a monitored line item — a number you will really check, a threshold that really matters, and an action you have already decided to take.

The pre-mortem is not just "think about the downside" dressed up. It exploits a specific, measured quirk of the mind. When researchers asked people to estimate why a future project might fail, they surfaced far fewer real risks than when they were told to imagine the project had failed and to explain the failure. Klein's collaborators called the second framing prospective hindsight, and in the studies it lifted people's ability to name concrete causes by roughly thirty per cent. The certainty in "it failed — now say why" gives the mind permission to go looking for reasons it would otherwise politely refuse to see.

That matters because of everything you have met earlier on this shelf. Before you own a stock, your excitement is quietly recruiting evidence for the case and waving away the evidence against it — the confirmation reflex, hiring rather than weighing. A plain question like "what are the risks?" runs straight into that reflex and comes back with a thin, reassuring list. The pre-mortem's grammar sidesteps it. By assuming the failure has happened, it removes the thing you are defending, and the inconvenient risks — the ones your hope was hiding — finally get named. It is disconfirmation made easy: instead of asking you to argue against something you love, it lets you start from a failure that has, in the exercise, already occurred.

There is a second reason to do it before you buy rather than after a fall, and it is one you have already seen do damage. After an outcome is known, memory rewrites itself; the cause that happened comes to feel obvious, and the ones that did not quietly vanish. A pre-mortem written after a 20% drop is contaminated by the drop — it narrows to whatever explains today's pain. Written in advance, while you still have nothing to defend and do not yet know how the story ends, it preserves your real uncertainty. That is the whole reason it is a pre-mortem and not a post-mortem: the honest version has to be captured before the result exists to bend it.

One caution, because the exercise is easy to misuse. The point is not to frighten yourself out of every idea, nor to generate the longest possible list of things that could go wrong. Almost anything could go wrong. The pre-mortem is a search for the few causes that would genuinely break this particular thesis — and, just as usefully, a way to notice when a risk is simply too large for the size you were planning.

Here is where most people's pre-mortem quietly fails: they do the imagining, feel suitably sobered, and stop. A page of fears is written, nodded at, and never looked at again. Nothing about the next results day is different. The fear was real and completely inert.

The repair is a small, mechanical discipline. Every imagined failure cause is pushed through four questions, and only a cause that survives all four becomes part of your defence:

1. The cause. What actually went wrong, in one plain sentence. "Customers stopped paying on time and the reported profit was never really cash."

2. The metric. Which single number, in a document you can actually reach, would show this happening early? For the cause above, it is receivable days — how long, on average, the company waits to be paid, read straight off the balance sheet. A cause with no metric is a mood; it cannot be monitored, so it is dropped or rewritten until it has one.

3. The threshold. At what level does the number stop being noise and start being the failure? Not "if receivables rise" — they wobble every quarter — but "if receivable days go above 90 for two consecutive quarters." A level, and a patience, so one bad print does not panic you and a real trend does not escape you.

4. The action. What will you do the day the threshold is crossed — decided now, while calm? Exit. Trim. Reassess. Writing the action in advance is what stops the frightened, later version of you from renegotiating it.

Run a cause through those four and it stops being a worry and becomes a monitoring trigger: cause → metric → threshold → action. That chain is the entire technique. It is the same machinery as the written thesis from the previous module — indeed a good pre-mortem is where field five of the thesis, the invalidation, actually comes from — but pointed at the whole cluster of ways the idea could die, not just the single headline break condition.

Notice what the chain quietly does beyond monitoring. If a cause is real but you cannot find any metric that would show it early, that is worth knowing — it is an unwatchable risk, and an unwatchable risk is an argument for a smaller position, not a bigger hope. This is where the pre-mortem hands straight to position sizing: the risks you cannot monitor, you size down for. The exercise does not only tell you what to watch; it tells you what you are flying blind on.

The method stays abstract until you see it done on a real company shape. Here is a pre-mortem for a made-up mid-cap manufacturer — every figure invented to teach the method, none a claim about any real business. The reader has imagined the position down 40% a year from now and asked, plainly, what did it: then pushed each answer through the four steps.

Read the second row, because it names a fact many beginners never think to watch. A promoter pledge is the founder borrowing money against his own shareholding in the company. A little is ordinary; a rising pledge is a warning, because if the lender ever forces a sale of those pledged shares the price can fall in a rush, and the person supposedly most aligned with you is the one under pressure. It is disclosed every quarter in the shareholding pattern, so "above 25%" is not a feeling — it is a number you can look up the same afternoon results are out. That is the test for every row: could a stranger, handed the filing, tell whether the line had tripped? If yes, it is a trigger. If only you can tell, by consulting your mood, it is a worry still wearing a rule's clothes.

Four rows is plenty. A watch-list you will actually read on results day beats an exhaustive one you will not. The discipline is not to catch every conceivable risk; it is to make the handful that would truly break the thesis impossible to ignore and impossible to argue away in the moment.

The fastest way to feel the difference between a worry and a trigger is to try to convert one. The builder below starts you with three plain fears — the sort a first pre-mortem produces. Your job is to give each a metric, a threshold, and an action, and watch the tool refuse to count any cause until all three are in. illustrative

It is deliberately stubborn about this. A page of eloquent fears scores nothing; a single cause with a number and an action becomes a line on your checklist. That is the whole lesson rendered mechanical: the pre-mortem is not the imagining, which is the easy and slightly thrilling part. It is the dull conversion of each imagined failure into something a colder, later version of you can read off a filing without having to feel brave.

Now put a finished checklist to work on the day it matters. Suppose your mid-cap reports, and the feed is loud — some cheering the revenue, some warning of a slowdown. Without the pre-mortem, that noise is all you have, and you read it with your mood. With it, results day becomes a short, calm errand: you open the accounts, read receivable days (68 — fine), the pledge (19% — fine), the debt ratio (2.4x — fine), and the plant date (on track). Nothing tripped. You close the tab. The pre-mortem did not make the quarter good; it made your reaction to it pre-decided.

Imagining failure well protects you from a great deal, and it is honest to mark its edges.

It does not make the thesis correct. You can run a flawless pre-mortem on a company you have simply misjudged, name four perfect triggers, and none of them fires while the business slowly disappoints in a way you never imagined. The exercise disciplines your attention; it does not supply insight into the business. What it guarantees is not that you were right, but that the specific failures you could foresee will be caught early and cleanly instead of late and expensively.

It cannot catch the failure you did not imagine. A pre-mortem only monitors the causes you thought of. The genuinely novel blow — a regulation nobody expected, a fraud below the numbers you can see — will not be on your list, which is one more reason the unmonitorable risks argue for a modest size rather than a confident one.

And it is worthless if the triggers are theatre. A threshold set so loose it could never trip, or a metric you never actually open on results day, gives the comfort of a defence without the defence. Worse, it can breed false calm — "I have a pre-mortem" becoming a reason to size up. The test is behavioural, not literary: on the next results day, did you really read those four numbers, and would you really have acted if one had crossed?

The same few slips turn a real pre-mortem back into a comforting ritual. Named once, they are easy to catch on the page.

1. Stopping at the fears. The vivid list feels like the work, and it is only the setup. A cause with no metric and no threshold never survives contact with results day. If it cannot be turned into a number, rewrite it until it can, or admit it is unmonitorable and size down.

2. Vague thresholds. "If margins fall" or "if debt rises" trips on every ordinary wobble and so, in practice, is ignored. A usable threshold has a level and a patience — below 22% for two quarters — so noise does not fire it and a real trend cannot escape it.

3. Forgetting the action. A trigger with no pre-decided response leaves the hardest choice for the worst moment. Write "exit", "trim", or "reassess" now, while calm, so the frightened version of you is merely following instructions.

4. Doing it after the fall. A pre-mortem written once the price has dropped is a post-mortem in disguise — hindsight has already narrowed and flattered the list. The honest version has to be captured before you own the stock, while you still have nothing to defend.

5. Moving the line when it trips. Widening 90 days to 110 on the day receivables cross 90 is not flexibility; it is the confirmation reflex overruling your calm self. If a threshold was wrong, it was wrong on the day you set it — revise it deliberately on a quiet day, never under the sting of a loss.