The template. Three prompts, in order: (1) "It has been six months. The position is down 50% from where I bought it. What is the most plausible single story for how this happened?" (2) "What were the load-bearing assumptions in my thesis that this story tells me were wrong?" (3) "For each one, what could I have checked BEFORE entering — even cheaply — that would have surfaced the problem?" Write 1-2 sentences per prompt. The exercise takes 30-60 minutes if done seriously.
Worked example — Halcyon Logistics. Prompt 1: "Halcyon's two largest customers (a regional grocer and a national parts retailer) re-bid their freight contracts; Halcyon lost the parts retailer to a low-cost competitor and the grocer cut volume 30%. Revenue dropped 22% in two quarters; the high-margin lane mix evaporated; EBITDA margin fell from 18% to 11%. The stock went from $46 to $23." Prompt 2: "I was assuming Halcyon's customer relationships were stickier than freight industry norms because of the dedicated equipment story management told. I had no independent evidence for that stickiness — just the management quote." Prompt 3: "I could have checked publicly available contract-expiration disclosures in customer 10-Ks; I could have talked to two former operations managers via LinkedIn for ten dollars and an hour; I could have looked at Halcyon's revenue concentration disclosure to see how much was at-risk." Result: the analyst goes back and does items in prompt 3 before opening the position.
| Pre-mortem failure mode | What it looks like | The corrective |
|---|---|---|
| Overconfidence | "I cannot think of a plausible story for a 50% loss" | If a 50% loss has no story, the analyst has not stress-tested the thesis enough — work the prompt until a story emerges |
| Plausibility creep | "The story is implausible because it requires three independent bad things to happen" | Single-driver stories are easier to write and more common in reality; do not require simultaneous independent failures |
| Sunk-cost bias | "I have spent three weeks on this; I do not want to find a reason to kill it" | Recognize the bias by name; ask the senior reader to do the pre-mortem and compare |
| Confirmation-only stories | All failure stories are about things the analyst already worried about | Force a story about the most boring, mundane failure cause; that is where the surprises live |
Plausibility creep is the most insidious of the three failure modes. The analyst, asked to imagine the loss, gravitates toward elaborate scenarios with multiple moving parts ("a recession, AND the new product fails, AND the CEO leaves"). The compound probability of any specific elaborate story is low — but the cumulative probability of SOME single-driver failure is high. Single-driver stories are what actually happen; insist on them.
Sit with the ideas.
An analyst has spent three weeks building a thesis on Halcyon Logistics. She is about to recommend a 4% long position. Her senior reader asks her to write a pre-mortem: "Assume you own this at 4% and it drops 50% in six months. Write the story of how that happened." What is the strongest reason this exercise is worth the hour it takes?