{
  "abstract": "Outcome-predictive signals are often treated as promising optimization targets in reinforcement learning, but predictive validity and optimization utility are different claims. We expose this distinction in GuanDan, a multi-round imperfect-information game in which the implication of a round outcome depends on the public match state. Held-out analysis shows that state--gain interactions provide incremental predictive value beyond a constant signed-gain effect, establishing that match progress is predictively meaningful. We then use changes in a frozen match-win predictor, $ΔΦ$, as a controlled round-level supervision intervention in a matched DanZero-style Monte Carlo self-play protocol. The intervention can succeed in an individual lineage: a selected progress-augmented checkpoint substantially outperforms its matched Baseline on a locked evaluation. Yet the method-level result reverses this impression: none of five preregistered replication seeds is positive, the pooled ten-seed interval includes zero, and across-iteration effects repeatedly change sign without a common late-stage advantage. These results identify a predictive--optimization gap: under the studied learner, a signal can describe outcome-relevant progress while failing to provide a reliably beneficial optimization direction.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.29411",
  "author_total": 5,
  "authors": [
    {
      "name": "Qicheng Zou"
    },
    {
      "name": "Biao Yuan"
    },
    {
      "name": "Siyu Song"
    },
    {
      "name": "Wentao Liu"
    },
    {
      "name": "Xiaokun Wang"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/2n28ewzcwebdy.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.29411",
  "fulltext_url": null,
  "id": "2n28ewzcwebdy",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.29411",
  "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",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Artificial Intelligence in Games"
    ]
  ],
  "openalex": "W7223221053",
  "pdf_url": "https://arxiv.org/pdf/2610.29411",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-21",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/2n28ewzcwebdy.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:34:29Z",
      "native_id": "2610.29411",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.29411"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "The Predictive--Optimization Gap in GuanDan Self-Play",
  "title_text": "The Predictive--Optimization Gap in GuanDan Self-Play",
  "type": "preprint",
  "updated_at": "2026-10-30T12:34:29Z",
  "url": "https://research.draconitic.com/paper/2n28ewzcwebdy",
  "venue": "arXiv",
  "versions": []
}