Skip to content

Odds for, odds against and probability

Odds compare an event with its complement. Probability compares the event with the combined total.

On this page

Read the ratio as two weights

For odds for an event of a to b, the favourable weight is a and the unfavourable weight is b. Normalize by their sum. The weights need not be literal observed frequencies.

P(E)=aa+b,P(Ec)=ba+bP(E)=\frac{a}{a+b},\qquad P(E^c)=\frac{b}{a+b}

a and b are nonnegative weights and a + b must be positive.

Convert a probability back into odds

For a probability p strictly between zero and one, odds for the event are p to 1 − p, or p/(1 − p) to 1. Odds against reverse those two components. At p = 0 or p = 1, one of the components is zero; a finite ratio-to-one representation may then be unavailable.

Equivalent representations
ProbabilityOdds forOdds against
20%1 to 44 to 1
50%1 to 11 to 1
72%18 to 77 to 18

Avoid dividing by the unfavourable weight alone

For odds for of 2 to 9, dividing 2 by 9 gives an odds ratio of approximately 0.2222 to 1. It does not give the event probability. Dividing by 2 + 9 gives approximately 0.1818, the event’s share of the total weight.

Keep for and against attached to the event

A quotation can be numerically identical while referring to the opposite direction. “9 to 1 for default” and “9 to 1 against default” imply 90% and 10% default probabilities respectively. Omitting the direction changes the interpretation.

Further references