Outside View Base-Rate Forecasting
Anchor predictions in comparable outcomes before adding case details
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 99%
The inside view starts with the details of the case, combines them with personal judgment, and projects forward. The outside view first reframes the case as one member of a larger reference class and asks what happened to similar cases before. That historical distribution supplies an anchor, exposes a wider range of possible outcomes, and counters the tendency to overvalue cherished case-specific information. The forecaster then incorporates genuinely relevant details to place the current case within the distribution. Mauboussin does not present the choice as all inside or all outside: the improvement comes from thoughtfully combining individual analysis with base rates. The method applies to everyday plans as well as investing.
Origin
Mauboussin learned the inside-view versus outside-view distinction from Daniel Kahneman and calls base rates the idea he would teach his 18-year-old self.
Core principles
- 01Treat the current problem as one instance of a larger class
- 02Historical outcomes constrain plausible forecasts
- 03Case-specific information still matters after the base rate
- 04A thoughtful blend beats either view alone
How to run it
- 1
Specify the forecast
Define the outcome precisely enough to compare it with prior cases. Examples include project duration, renovation cost, company growth, or the probability of winning.
Pro tip Use a measurable outcome and time horizon.
Watch out A vague question produces an arbitrary reference class.
- 2
Find the reference class
Identify prior cases that faced materially similar conditions. Ask what happened when other people or organizations were in this situation.
Pro tip Start broad, then narrow only when the narrower class has enough observations.
Watch out Choosing comparables solely because they support your preferred answer recreates confirmation bias.
- 3
Map the distribution
Examine the average, range, and frequency of outcomes in the reference class. Preserve the full distribution rather than extracting only one headline number.
Pro tip Look at both the typical result and the tails.
Watch out Small samples can make precise-looking base rates unstable.
- 4
Anchor outside
Use the reference-class result as the starting estimate. Explicitly compare it with the forecast generated from case details alone.
Pro tip Write down both estimates before reconciling them.
Watch out Do not discard an inconvenient base rate because the current case feels special.
- 5
Adjust with evidence
Move the estimate within the historical distribution when case-specific evidence justifies it. Make each adjustment explicit and proportionate.
Pro tip Use measurable differences such as speed figures, size, industry, or demonstrated performance.
Watch out Narrative excitement is not evidence of exceptional odds.
- 6
Record the final range
Express the result as a range or probability that acknowledges uncertainty. Revisit it as new information arrives.
Pro tip Track forecast outcomes to improve future reference-class choices.
Watch out A single point estimate can hide overconfidence.
In the wild
Big Brown entered the Belmont with market odds implying a 77% chance of completing the Triple Crown. Yet only about 40% of all prior contenders had succeeded, and only three of 20 since 1950 had done so. The horse also had the slowest speed figure among the last seven contenders, all of which had failed.
→ The outside view indicated a much lower probability, and Big Brown finished last.
Instead of assuming a company will grow between 2% and 8%, an investor can examine the historical growth distribution of companies of similar size and in similar industries. That distribution may reveal a much wider set of plausible outcomes before the investor places the company within it.
→ The forecast reflects both historical variation and company-specific analysis.
Common mistakes
Declaring the case unique
People naturally overweight their own information and can dismiss comparable outcomes without sufficient evidence.
Choosing the wrong class
A reference class that is too broad, too narrow, or selected after seeing the answer can mislead rather than calibrate.
Replacing analysis with a base rate
Mauboussin recommends a thoughtful combination of outside and inside views, not the mechanical use of history alone.
Is it for you?
Best for
It is best for estimating costs, completion times, growth rates, success probabilities, and other outcomes with historical analogues.
Not ideal for
It is not ideal when no meaningful reference class exists or the process has changed so radically that old outcomes are uninformative.
From the transcript
“The outside view or the base rates by contrast says, I'm going to think about my problem as an instance of a larger reference class.”
“I'm going to ask a really simple question like what happened when other people were in this situation before”
“some thoughtful combination of the inside versus the outside view. Sort of your own analysis versus base rates tends to lead to better and more…”
From the episode
#659: Michael Mauboussin — How Great Investors Make Decisions, Harnessing The Wisdom (vs. Madness) of Crowds, Lessons from Race Horses, and More
Michael Mauboussin