Diebold-Mariano test converges to non-Gaussian stable limits under infinite-variance loss differentials, causing severe size distortions, with sub-sampling proposed as valid inference independent of tail index.
Stochastic Processes and their Applications , volume=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Heavy Tails and Predictive Ability Testing
Diebold-Mariano test converges to non-Gaussian stable limits under infinite-variance loss differentials, causing severe size distortions, with sub-sampling proposed as valid inference independent of tail index.