A forecast may assign probabilities to discrete events or describe a distribution over numerical values. The target, time horizon and relevant conditions must be clear for those probabilities to be interpretable.
Quality cannot be established from one outcome alone. Repeated forecasts and suitable evaluation are needed to judge whether the assigned probabilities are informative and consistent with observed frequencies.
ELI5
Probabilistic forecasting gives chances for different outcomes instead of pretending that one result is certain. A less likely outcome can still happen.
For example, a weather forecast might give rain a 30 percent chance. One rainy day does not by itself show that the forecast was wrong.
Is the most likely outcome guaranteed to occur?
No. A forecast can assign the highest probability to one outcome while still allowing substantial probability for alternatives.
Does a detailed distribution prove a forecast is reliable?
No. The estimates still depend on evidence and modelling assumptions and need evaluation against outcomes.
