Base Rates: Start With the Odds
Before you judge a specific case, ask how often things like it turn out each way. That baseline is the best first guess you have.
A founder pitches you: “our startup will be the next billion-dollar company — the team is brilliant and the product is magic.” Your gut wants to judge it on the vivid story in front of you. But there’s a quieter, far more accurate way to forecast that professional predictors reach for first, and it starts by ignoring the story for a moment. It’s the highest-leverage forecasting habit there is, and almost nobody uses it. Hold the question: what should you look at before the specifics of any particular case?
The base rate is how often it happens in general
The base rate of an outcome is simply how frequently that outcome happens across all similar cases. Most new restaurants close within a few years; most startups don’t become unicorns; most drivers think they’re above average. Before you look at any specifics, the base rate is your best starting estimate — the odds for a random member of the group. This is the “outside view”: judging a case by the track record of its reference class (the group it belongs to), not by its own shiny details. It’s the numerical cousin of separating signal from noise (lesson 2): the base rate is durable signal; the pitch is often noise.
The inside view feels smarter and misleads
Why doesn’t everyone do this? Because the inside view — reasoning from the specific story, the “this time is different” details — feels more intelligent and is far more vivid. A great team and a magic demo scream “winner,” drowning out the boring fact that 99% of similar bets lose. Vividness hijacks judgment. The fix isn’t to ignore the specifics; it’s to anchor on the base rate first, then adjust for genuinely unusual features — and to be honest that most cases aren’t as special as they feel from the inside.
Two ways to forecast a startup: • Inside view: “Brilliant founders, huge market, slick product → this will be massive.” Confident, vivid, and usually wrong. • Outside view: “Roughly 1 in 100+ funded startups reaches a billion dollars. Start there. Is this one genuinely, evidence-backed exceptional? Maybe nudge to 1 in 50 — still likely no.” Less exciting, far better calibrated.
Anchor, then adjust — carefully
The full method: (1) find the right reference class (“funded startups in this sector,” not “companies in general”), (2) take its base rate as your anchor, (3) adjust up or down only for features with real predictive weight — and adjust less than your gut wants, because the inside view over-weights specifics. Choosing the reference class is where judgment lives: too broad and it’s meaningless, too narrow and you have no data. Done well, base-rate thinking quietly beats elaborate case-by-case reasoning across almost every domain — medicine, investing, project timelines, and, yes, forecasting the future.
Imagine guessing how long a home renovation will take. The inside view walks the rooms, imagines each task going smoothly, and says “six weeks.” The outside view asks the contractor, “of the last fifty jobs like this, how long did they actually take?” — and the honest answer is “four months, because something always goes sideways.” The inside view is a hopeful story; the base rate is the scoreboard of what really happens. When they disagree, bet on the scoreboard and only nudge it for what’s genuinely special about your house.
Using a base rate to cut through hype: 1. Claim: “This new gadget category will be in every home within two years.” 2. Reference class: past “revolutionary” consumer gadgets. Base rate of two-year mass adoption? Very low — most take far longer or fizzle (recall Amara’s law, lesson 4). 3. Anchor there: probably not two years. Now adjust for specifics — is distribution unusually easy, the price unusually low? Maybe nudge slightly. 4. Your forecast lands near “unlikely in two years, plausible in five-plus” — grounded, not swept up in the launch excitement.
This is the reading. The interactive version — active-recall quiz, a hands-on experiment you run in your own AI, and an earned mastery check — is free in the app.
Start this lesson free →