What it means
Every forecast starts somewhere. Before the test results, the earnings report, or the market open, you hold a belief about the odds, and that starting belief is the prior probability.
Bayesian thinking makes the starting point explicit and the updating mechanical: the prior combines with the strength of new evidence to produce the posterior, the improved estimate. Bayes' theorem is the engine.
It says the probability of a hypothesis given evidence depends on how likely the evidence is under the hypothesis, weighted by the prior, relative to how likely the evidence is overall. The prior's power is easiest to see in medical testing.
A disease affecting one person in a thousand, tested with a test that is 99 percent accurate, still leaves most positive results false, because the tiny prior dominates the arithmetic. Finance runs on the same trap.
A fraud screen flagging one percent of firms as suspicious means most flagged firms are innocent, and analysts who ignore the base rate waste weeks chasing ghosts. Where priors come from is the honest weakness of the method.
Sometimes they are measured base rates; sometimes they are judgement, and two analysts with different priors can see the same evidence and reach different posteriors. That subjectivity is also the method's honesty: it forces the starting assumption into the open where it can be argued with, instead of hiding it inside a confident conclusion.
For a non-finance reader, the habit is one question: before this new information arrived, how common was the thing I'm now worried about? The answer disciplines every headline, test, and hot tip.
Courts meet the same arithmetic in forensic evidence. A rare DNA profile means little until the size of the suspect pool enters the calculation, which is why statisticians insist on presenting base rates alongside match probabilities.
Hiring and admissions decisions benefit too. A prestigious signal loses its shine when the prior success rate of its holders is examined, and structured interviews beat intuition partly because they anchor on base rates rather than impressions.
In practice
Real-world examples.
Example
An analyst sets a 5% prior for default in a loan pool, then updates it upward as late payments arrive. The prior did the heavy lifting before a single payment was late, and each new batch of repayment data moves the estimate again.
Example
A doctor ignores the disease's rarity and tells a patient that a positive screening means cancer is almost certain, the classic base rate error. A short conversation about how common the disease is in people with no symptoms would have changed the message completely.
Example
A fund updates its prior on a strategy working, from 60% down to 30%, as three flat quarters of evidence arrive. The manager records the change in a memo so that the investment committee can see exactly which evidence moved the number and by how much.
Formula
Calculation
Bayes' theorem: posterior probability = (likelihood of the evidence given the hypothesis x prior probability) / total probability of the evidence.
Worked example. A screening test is used on a group where 1% have a condition. The test is 99% sensitive (it catches 99% of true cases) and 95% specific (it correctly clears 95% of healthy cases). Take 100,000 people: 1,000 have the condition and the test flags 990 of them (1,000 x 99%). Of the 99,000 healthy people, the test wrongly flags 4,950 (99,000 x 5%). Total positives are 990 + 4,950 = 5,940, so the chance that a positive result is real is 990 / 5,940 = about 16.7%, roughly 17%. The small 1% prior is why most positives are false alarms despite an apparently excellent test.Case study
Seen in the real world.
This case study is fictional and illustrative. A made-up bank's compliance team deploys a transaction screen that flags possible money laundering with 95 percent accuracy. In month one it flags 400 accounts, and the team celebrates, until the statistician asks for the base rate: laundering occurs in perhaps one account per five thousand. The arithmetic sobers the room.
With a prior of 0.02 percent, even a 95 percent accurate flag means the vast majority of flagged accounts are innocent, and the posterior probability that any single flag is real is under one percent. The team redesigns its workflow: the screen becomes a first pass, and flagged accounts are scored against independent risk factors before any filing. Review time falls, true positives triple, and the statistician prints the Bayes calculation on the wall, with the prior circled, as the team's permanent reminder that evidence never speaks without context.
Watch out
Common mistakes.
- Ignoring the base rate; a strong test applied to a rare condition still produces mostly false alarms.
- Treating the prior as objective truth; it is a stated starting assumption, and honest analysis says where it came from.
- Updating only once; evidence arrives in streams, and each piece should move the probability again rather than resetting the debate. The stream never stops moving the number.
Questions
People also ask.
What is a prior probability?
The estimated probability of a hypothesis before new evidence is considered, which Bayes' theorem updates into the posterior probability.
Why do priors matter so much?
Because evidence is interpreted against background rates: rare events stay unlikely even after seemingly strong signals, unless the evidence is overwhelming.
Where do priors come from?
Measured base rates when available, otherwise structured judgement; the method's discipline is stating the prior openly so it can be challenged.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%