Skip to content

Combining conditional probabilities with scenario weights

Scenario weights of 40%, 35% and 25%, with conditional positive-return probabilities of 20%, 55% and 85%, give an unconditional positive-return probability of 48.5%.

On this page

Use a complete, non-overlapping scenario set

The fictional rate model has rise, unchanged and fall scenarios. Their weights sum to one. A bond fund’s positive-return probability is specified separately within each scenario. The numbers are learning assumptions, not estimates of actual future interest rates or fund performance.

Calculate each scenario’s contribution

Multiply each conditional probability by the probability of its scenario. The contributions are joint probabilities of the scenario and positive return.

Weighted branches
ScenarioScenario weightPositive return given scenarioJoint positive-return contribution
Rise40%20%8%
Unchanged35%55%19.25%
Fall25%85%21.25%
Total100%48.5%

Explain the unweighted-average error

A simple average of 20%, 55% and 85% gives approximately 53.33%. That calculation gives equal weight to the three scenarios, contrary to the supplied 40%, 35% and 25% weights. It answers a different model.

Do not reuse a branch probability in the other direction

Under these assumptions, the probability of a rise given positive return is 8%/48.5%, approximately 16.49%. It is not the 20% probability of positive return given a rise. The conditioning event changed, so the denominator changed.

Further references