Activate this model whenever you are evaluating a potential opportunity (e.g., business, investment, career, or personal).
Work through these checkpoints in order. A red flag at any stage warrants careful scrutiny before proceeding.
Signal/Noise: Is this opportunity the result of genuine insight, or pattern-matching on randomness? The opportunity feels compelling but you can’t articulate why precisely.
Asymmetry: Is the upside meaningfully larger than the downside? Upside requires many things to go right; downside requires only one thing to go wrong.
Reversibility: Can you exit gracefully if you’re wrong? What does exit cost? Exit costs are non-obvious or understated.
Skill vs. Luck: How much of success here is within your control? Success is attributed to skill but is actually highly luck-dependent.
Timing: Is this the right idea at the right time? Others have tried and failed without clear explanation of what’s different now.
Second-Order Effects: What happens if it succeeds? Does success change the landscape in ways that eliminate the advantage? The opportunity disappears if it works at scale.
Ruin Check: What is the worst realistic outcome? Is it survivable? The worst case is ruin or near-ruin.
Before evaluating the specific opportunity, find the base rate for its category:
What fraction of opportunities like this have succeeded historically?
What do they look like when they fail?
What do the successful ones have in common that this one does or does not share?
Your enthusiasm for the specific opportunity is not evidence that it is above average. Anchor on the base rate and require specific, credible evidence to justify upward adjustment.
Idea quality and timing are independent variables. Many good ideas fail because of timing; many mediocre ideas succeed because of it.
Have others tried this before? If yes and failed, what is structurally different now?
What trends does this require? Are those trends actually in motion, or hoped-for?
Why now rather than two years ago or two years from now? Forces a precise answer, not a vague one