One desired property of probabilistic forecasts is that forecasted event probabilities $f$ coincide with observed relative frequencies $F$. This is visualized in reliability plots that plot the latter variable versus the former,

\begin{equation} F_i = \dfrac{\sum o|f=i}{J_i} \end{equation}

where $F$ is the observed relative frequency, $i$ is one of eleven allowable values of the forecast $f$: $i~\in~\lbrace~0,0.1,0.2,\ldots,1~\rbrace$, $o$ is an indicator variable that is assigned a value of 1 if the event was observed and a value of 0 if the event was not observed, $f$ is the forecasted probability of event occurrence and $J_i$ is the total number of forecasts made within bin $i$ [wilks_statistical_2011].