{
  "abstract": "Changing a deployment policy can invalidate a world model's uncertainty while leaving the physical transition mechanism and predictor unchanged. For one-step prediction, if source and target laws share the conditional mechanism $P(Y|X)=Q(Y|X)$, then the intervention is identified by the pre-outcome change of measure $w(X)=dQ_X/dP_X$. For policy-controlled dynamics this ratio factors into state-occupancy and action-propensity terms. Weighted conformal calibration transports marginal coverage to the target law; with an estimated ratio $\\hat w$, $P_Q\\{Y \\in C_{\\hat w}(X)\\} \\ge 1-α-TV(Q,\\hat Q) \\ge 1-α-E_P|w-\\hat w|$. Population log-loss regret for a source-versus-target domain classifier bounds the same ratio error. If $h$ is its clipped target posterior, $ρ$ is the target-domain prior, and $Δ(h)$ is population log-loss regret, then $E_P|w-\\hat w_h| \\le (1/(ργ))\\sqrt{Δ(h)/2}$. An exact cancellation improves the dependence from a generic odds-map bound of order $γ^{-2}$ to $(ργ)^{-1}$. Our experiments use a released DINO-feature world model and physical Franka trajectories. Under the strongest controlled state-action shift, ordinary 90% conformal sets cover 55.0%, whereas oracle and learned-ratio calibration cover 91.0% and 90.9%. The learned surrogate remains within 0.9% total variation of the target law. Calibration is relative to the deployment law: the model can remain fixed while calibration follows the intervention. We machine-check in Lean 4 the finite-support algebraic and causal identities used in the classifier-regret argument, including its coverage-composition step.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.35877",
  "author_total": 2,
  "authors": [
    {
      "name": "Ayush Mittal"
    },
    {
      "name": "Dhruv Gupta"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/cpw6w1e4tywqd.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.35877",
  "fulltext_url": null,
  "id": "cpw6w1e4tywqd",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.35877",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "cs.RO"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Reinforcement Learning in Robotics"
    ]
  ],
  "openalex": "W7224118433",
  "pdf_url": "https://arxiv.org/pdf/2610.35877",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-01",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/cpw6w1e4tywqd.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:35:59Z",
      "native_id": "2610.35877",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.35877"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "Calibrated for Which Policy? Conformal Uncertainty for Fixed World Models Under Intervention",
  "title_text": "Calibrated for Which Policy? Conformal Uncertainty for Fixed World Models Under Intervention",
  "type": "preprint",
  "updated_at": "2026-10-30T12:35:59Z",
  "url": "https://research.draconitic.com/paper/cpw6w1e4tywqd",
  "venue": "arXiv",
  "versions": []
}