{
  "as_of": "2026-10-19",
  "question": "Does a separately fitted, single fundamentals margin add predictive information beyond the existing Senate polling forecast?",
  "primary_test_cycles": [
    2016,
    2018,
    2020,
    2022,
    2024
  ],
  "extended_sensitivity_cycles": [
    2012,
    2014,
    2016,
    2018,
    2020,
    2022,
    2024
  ],
  "specifications": {
    "lean_incumbency": [
      "pvi_scaled",
      "incumbency_advantage"
    ],
    "candidate_only": [
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled"
    ],
    "candidate_lean": [
      "pvi_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled"
    ],
    "candidate_lean_strict": [
      "pvi_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "strict_news_scaled"
    ],
    "lean_incumbency_generic": [
      "pvi_scaled",
      "incumbency_advantage",
      "generic_scaled"
    ],
    "candidate_lean_generic": [
      "pvi_scaled",
      "generic_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled"
    ],
    "finance_core": [
      "pvi_scaled",
      "incumbency_advantage"
    ],
    "candidate_money": [
      "pvi_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled",
      "individual_share_difference"
    ],
    "candidate_money_generic": [
      "pvi_scaled",
      "generic_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled",
      "individual_share_difference"
    ],
    "candidate_money_amount": [
      "pvi_scaled",
      "incumbency_advantage",
      "incumbent_prior_scaled",
      "incumbent_prior_missing",
      "nonincumbent_prior_scaled",
      "nonincumbent_prior_missing",
      "nonincumbent_experience",
      "career_missing_advantage",
      "appointed_advantage",
      "multiple_elections_advantage",
      "news_scaled",
      "individual_gap_10m"
    ]
  },
  "index": "Ridge prediction of actual D-minus-R margin using no current race polling as an input. Coefficients use only earlier elections; prior personal performance also predates the target election.",
  "history": "218 polled historical races, 2006-2024, train pure indices. The frozen 2010-2024 baseline supplies paired tests; the primary 2016-2024 cohort has 130 races.",
  "features": "Incumbency; appointed vs multiple election wins; latest comparable prior personal overperformance for incumbent and nonincumbent; nonincumbent state/House/governor experience hierarchy; dated public-news severity; optionally partisan lean, cutoff-dated generic ballot, and Q3 individual-dollar share or amount.",
  "missing": "Source and prior-performance gaps have explicit signed missing indicators; financial and generic-ballot gaps are excluded from their complete-case specifications, never zero-imputed.",
  "ridge_multipliers": [
    0.25,
    1.0,
    4.0
  ],
  "blends": [
    0.0,
    0.1,
    0.25,
    0.5
  ],
  "blend_formula": "P + clip(alpha * (F - P), -3, +3) percentage points. F is the one-number fundamentals estimate and P is the frozen active forecast.",
  "selection": "For each outer election, fit and score inner indices using only still-earlier elections. Select ridge strength and blend on the mean MAE across at least two earlier inner validation elections. With fewer, use lambda 1 and alpha 0. Also show fixed 25% blends as exploratory diagnostics, with lambda 1.",
  "training": "Equal-cycle race weights with an eight-year recency half-life. Intercept penalty .10; PVI/10 prior 10, penalty .10; incumbency prior zero, penalty .5; added features prior zero, penalty 2; multiplied by lambda. No current-cycle outcome or race polling enters the pure index.",
  "primary_comparison": "Candidate + lean nested blend versus frozen active forecast, and versus a separately nested lean/incumbency blend on identical races. Other variants are sensitivities, not a search for an unadjusted winning p-value.",
  "inference": "Two-sided exact sign flips of whole-election paired MAE gains, plus exploratory 50,000 cycle-bootstrap intervals. Holm adjust all available primary-versus-active p-values. The five-cycle minimum attainable two-sided exact p is .0625, assuming nonzero differences. Financial complete cases have one supported later test election; no significance claim.",
  "probability": "Keep inherited per-cycle uncertainty/calibration fixed for paired diagnostics. A point-error gain alone does not validate uncertainty or authorize adoption.",
  "production_changed": false,
  "exploratory": true
}