Workshop
Learn · ForecastingExplainer

Base Rate Fallacy Explained, With a Worked Example

Workshop · Written by Workshop. Examples come from the live record.Updated November 21, 2026
Short answer

The base rate fallacy is the mistake of ignoring how common something is in general (its base rate) when judging a specific case, and leaning on the vivid details of the case instead. A positive result from an accurate test for a rare condition is more likely to be a false alarm than a true one, because the healthy people far outnumber the sick.

The classic example: a screening test

Take a condition that 1 in 100 people have. A test catches 90% of the people who have it, and wrongly flags 5% of the people who do not. You test 10,000 people and one comes back positive. How likely is it that this person has the condition? Most people say around 90%. Work it through:

People with the condition: 1% of 10,000 = 100
Correctly flagged: 90% of 100 = 90
People without it: 10,000 - 100 = 9,900
Wrongly flagged: 5% of 9,900 = 495
All positives: 90 + 495 = 585
Chance a positive is real: 90 / 585 = about 15%

About 15%, not 90%. The test is good, but the condition is rare, so false alarms from the many healthy people swamp the true hits from the few sick ones. The numbers are illustrative, not Workshop data. The error is treating the test's accuracy as the answer and forgetting the base rate, the 1 in 100.

Why it matters when forecasting

Forecasts invite the same mistake. A vivid story arrives, such as a strong product launch or a hot economic indicator, and the forecaster prices the story and forgets how often outcomes of this kind happen at all. If a company usually reports in line or better, a good-sounding reason to expect a miss still starts from a low prior. The right habit is the one in the example above: begin with the base rate, then adjust for what is specific to the case, and adjust in proportion to how strong the evidence really is.

There is a second way to commit it: using the wrong base rate. The share of all companies that miss is not the share for this company, and the share of months jobs rose in a boom is not the share in a recession. Choosing the right comparison group is the hard part, and it is covered in what a base rate is.

How Workshop's baseline is the defence

Every Workshop call carries a baseline, computed by code from comparable past cases, using only what was public before the call was registered. The forecast is scored against that baseline with the Brier score, and copying the baseline scores exactly zero. A forecast earns credit only by adding something the history did not already say, so a forecaster that ignores base rates cannot look good by telling a persuasive story.

A live example:

Cisco Earnings (Nov 18): Will It Miss the $1.35 Estimate? We Say 10% The baseline is 12%. How it was computed, as recorded with the call: base rate, K=4 own reports at the cent shrunk toward the pooled S&P 100 rate (75/425, m=8).

The two recipes are written out in what a base rate is. You can see the scored results, whatever they are, on the graded record.

How to avoid it

  • Ask first how often this kind of thing happens, before asking about the details of this case.
  • Pick a reference group that is neither so narrow it has three members nor so broad it has nothing in common.
  • Turn the evidence into a likelihood: how much more often would I see this if the thing were true than if it were not?
  • Write the number down before the answer is known, so it can be checked.

Common questions

What is the base rate fallacy in simple terms?

It is ignoring how common something is overall when judging one case. If a rare condition affects 1 in 100 people, even a fairly accurate test will produce mostly false alarms among the positives.

Is base rate neglect the same as the base rate fallacy?

Yes. Both names describe the same error of underweighting the general frequency in favour of case-specific details.

What is an example of the base rate fallacy in investing?

Expecting a dramatic outcome because of a compelling story without asking how often outcomes like that happen for comparable companies or data releases. Workshop's baseline exists to make that question unavoidable: see the graded record.

Live calls on this topic

Keep reading