Proper-score observation-driven filters: local geometry, estimation, and continuous-time limits
Abstract
Observation-driven filters typically use likelihood-score updates, corresponding to the logarithmic scoring rule. We generalise these updates to negative parameter derivatives of differentiable proper scoring rules, under a declared working family and predictable scaling. The rule determines the conditional risk projection and tail response, scaling converts its derivative into the update, and the autoregressive component determines the composite dynamic centre. Locally, risk curvature and scaling govern mean reversion, while the variance of the scaled innovation governs update noise. Under correct specification, unscaled curvature and score variance coincide for the log score by the information identity, but generally differ for other proper rules. For static parameters, we establish consistency and asymptotic normality under explicit stability and fixed-tuning conditions. In high-frequency scale models, centred updates converge to diffusions, whereas non-centred updates follow deterministic mean flows. For time-varying rule-specific projections, the local tracking error admits an Ornstein-Uhlenbeck approximation up to a stopping time, allowing correlated update and target shocks. Simulations and an international-equity application illustrate how criterion choice affects robustness, adaptation, variance-forecast loss, value-at-risk calibration and probability-integral-transform diagnostics. Predictive density and updating criterion are distinct design choices whose relative performance depends on the target and disturbance.
Related
- Learning Probabilistic Filters with Strictly Proper Scoring Rules
- On Solving Continuous-Discrete Projection Filters via Sum-of-Squares Relaxation
- A paradoxical route from hyperbolic geometry to proper proximality
- Finite Weil Jets, Identity Transfer, and the Exponential Geometry of Proper Orthogonal Tensors
- SN-ASMO: Satellite-Navigation Array Spatial-Manifold Precise Observation Theory A Mathematical Foundation for Observation Formation, Unified U(1) Geometry, Intrinsic Information, and Preservation of the RTK Integer Structure
BibTeX
@misc{livieri2026properscore,
title = {{Proper-score observation-driven filters: local geometry, estimation, and continuous-time limits}},
author = {Giulia Livieri and Gianluca Palmari},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2608.02828},
eprint = {2608.02828},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.02828},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Proper-score observation-driven filters: local geometry, estimation, and continuous-time limits
AU - Giulia Livieri
AU - Gianluca Palmari
PY - 2026
DA - 2026/09/21
PB - arXiv
DO - 10.48550/arxiv.2608.02828
UR - https://arxiv.org/abs/2608.02828
AB - Observation-driven filters typically use likelihood-score updates, corresponding to the logarithmic scoring rule. We generalise these updates to negative parameter derivatives of differentiable proper scoring rules, under a declared working family and predictable scaling. The rule determines the conditional risk projection and tail response, scaling converts its derivative into the update, and the autoregressive component determines the composite dynamic centre. Locally, risk curvature and scaling govern mean reversion, while the variance of the scaled innovation governs update noise. Under correct specification, unscaled curvature and score variance coincide for the log score by the information identity, but generally differ for other proper rules. For static parameters, we establish consistency and asymptotic normality under explicit stability and fixed-tuning conditions. In high-frequency scale models, centred updates converge to diffusions, whereas non-centred updates follow deterministic mean flows. For time-varying rule-specific projections, the local tracking error admits an Ornstein-Uhlenbeck approximation up to a stopping time, allowing correlated update and target shocks. Simulations and an international-equity application illustrate how criterion choice affects robustness, adaptation, variance-forecast loss, value-at-risk calibration and probability-integral-transform diagnostics. Predictive density and updating criterion are distinct design choices whose relative performance depends on the target and disturbance.
ER -