6533b822fe1ef96bd127d8d0
RESEARCH PRODUCT
Predictive distributions that mimic frequencies over a restricted subdomain
Giuseppe SanfilippoFrank Ladsubject
A_n and H_n distributionSequenceSettore MAT/06 - Probabilita' E Statistica MatematicaComputer scienceConditional probabilityInferenceFrequencyOutcome (probability)Reduction (complexity)Distribution (mathematics)Settore SECS-S/06 -Metodi Mat. dell'Economia e d. Scienze Attuariali e Finanz.Probability elicitationExtendable exchangeabilityFeature (machine learning)Probability boundSettore SECS-S/01 - StatisticaGeneral Economics Econometrics and FinanceAlgorithmFinanceEvent (probability theory)description
A predictive distribution over a sequence of $$N+1$$ events is said to be “frequency mimicking” whenever the probability for the final event conditioned on the outcome of the first N events equals the relative frequency of successes among them. Exchangeable distributions that exhibit this feature universally are known to have several annoying concomitant properties. We motivate frequency mimicking assertions over a limited subdomain in practical problems of finite inference, and we identify their computable coherent implications. We provide some examples using reference distributions, and we introduce computational software to generate any complete specification desired. Theorems on reduction and extendability of frequency mimicking assertions delineate the extent of the usefulness of such distributions.
| year | journal | country | edition | language |
|---|---|---|---|---|
| 2020-05-11 | Decisions in Economics and Finance |