Solar flares mostly originate from active regions, with more complex sunspot structures and magnetic polarity configurations producing more frequent and larger magnitude flares. Average flaring rates are commonly used to calculate Poisson probabilities for flaring in 24-hour periods, but these rates are typically determined from active region classification schemes with very broad parameter classes. Here we present improved separation of average flaring rates through the combination of multiple active region classification schemes and ranking of their potential prediction performance by the application of appropriate forecast verification metrics. The intended implementation of this novel technique will be presented in the broader context of the H2020 Space Weather “Flare Likelihood and Region Eruption Forecasting” (FLARECAST; http://flarecast.eu) project that launched in January 2015.