Base rates: the forecaster's first move

Ask someone whether a specific startup will still exist in five years and they will usually start telling you about the founder, the product, the market. All of that matters — but it is the wrong place to begin. The better first question is boring: of all companies that looked like this one, how many were still around after five years? That number is the base rate, and reaching for it first is the habit that most reliably improves a forecast.

The outside view versus the inside view

Psychologist Philip Tetlock, who studied thousands of forecasts in his research on so-called superforecasters, describes two ways to look at any question. The inside view builds a story from the specific details in front of you. The outside view treats the question as one instance of a broader class and starts from how that class usually turns out.

The inside view feels more intelligent because it engages with the specifics. That is exactly why it misleads. A vivid, detailed narrative is persuasive whether or not it is likely, and our minds reward the story over the statistics. Tetlock's finding was blunt: the forecasters who did best anchored on the outside view first, then adjusted. As a group, his superforecasters were roughly 30% more accurate than the professional analysts they were benchmarked against.

The order matters. Start with the base rate, then let the specifics move you up or down from it. Start with the story and the base rate rarely gets a look in.

What a base rate looks like in practice

Suppose you want to forecast whether a sitting government wins re-election. The inside view pulls you into this month's scandals and speeches. The outside view asks: across recent elections in comparable democracies, how often did the incumbent win? If the answer is roughly half, then "this time" needs a real, specific reason before you stray far from 50%.

The same move works on smaller questions. Will a particular building project finish on schedule? Large infrastructure projects overrun far more often than not, so "on time" should start as the underdog. Will an away team win a league match? Home advantage is a well-documented base rate; the away side starts behind before you know anything about the line-ups.

None of this means the details do not matter. It means the details are an adjustment, not a starting point. A strong base rate keeps you honest when a compelling story is trying to drag your estimate somewhere it should not go.

Why prediction markets are base rates in motion

A prediction market price is, in effect, a crowd's blended estimate — part base rate, part fresh information — expressed as a single number. When a market on Outcomer shows a Yes share at 30¢, it is telling you the crowd's all-in probability is about 30%. If you have done your own outside-view work, that price gives you something to test yourself against.

This is also why markets are worth watching over time rather than glancing at once. Prices drift as new information arrives, and the size and speed of a move tells you how much the crowd thinks a piece of news actually changed the base rate. A number that jumps on a headline and then settles back is a market that decided the story did not really move the odds. We wrote more about how those numbers behave in reading the odds, and about why the crowd's blend often beats a lone expert in the wisdom of crowds.

To take a current example: as of early August 2026, markets on major platforms were pricing a high probability — in the high 80s percent — that the U.S. Federal Reserve makes zero rate cuts across the whole of 2026, after the Fed held its benchmark steady at its late-July meeting. Whether or not that proves right, it is a clean illustration of a base rate in motion: a starting expectation, continuously repriced as each new data release either confirms or challenges it.

How to use base rates without overreaching

Base rates are a starting point, not a verdict. Three habits keep them useful. First, define the reference class honestly — "startups like this one" is only helpful if you are not quietly cherry-picking the comparison to fit the answer you already want. Second, adjust for genuinely distinctive features, but adjust in small steps; a truly unique case is rarer than it feels. Third, write the number down before you read the market, so you can see where you and the crowd disagree and ask which of you has the better reason.

Doing this well is a skill, and like any skill it improves with reps and honest scorekeeping. That is exactly what practice markets are for. On Outcomer you can form an outside-view estimate, place it against the crowd's price with virtual money, and watch how events actually resolve — no real cash at risk while you learn where your base rates are sharp and where they need work. It is the cheapest way to find out whether your forecasting instincts are as good as your stories.