Finite-horizon quantile martingale posteriors: raw-urn laws and matrix-gain regression
Martingale posteriors quantify uncertainty by forward-imputing observations from one-step-ahead predictive distributions, but implementations stop after finitely many imputations. For the empirical Pólya-urn posterior of a quantile the law of the stopped state is derived. The quantile of the stopped urn measure keeps the familiar martingale tail-sum variance fraction; the deployed stochastic-approximation tracker with frozen gain $c$ does not. Its variance carries an explicit factor $G_a$ with $a=cf_0(q_τ)$, which may fall below or exceed the tail fraction, and a density-adapted gain…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2607.19839
curl -X POST /api/v1/records/arxiv/2607.19839/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2607.19839/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'