{
  "task_id": "F6-conf-02",
  "sample_start": "2010-01-04",
  "sample_end": "2019-12-31",
  "n_treated": 0,
  "baseline_metric": -0.042851,
  "model_metric": null,
  "effect_estimate": null,
  "effect_ci_low": null,
  "effect_ci_high": null,
  "conclusion": "premise_invalid",
  "feasibility": "infeasible",
  "feasibility_reason": "data/gc5y_daily.csv ends on 2019-12-31 and contains zero rows with post_cut = 1; the March-2020 post-cut window and the 120-day window preceding it are absent, so the estimand cannot be computed.",
  "data_issues": [
    {
      "issue": "Sample covers 2010-01-04 to 2019-12-31 only (the Valet request used end_date=2019-12-31); no row has post_cut = 1, so there are zero post-cut trading days and the event (2020-03-04) lies outside the file.",
      "handling": "Effect estimand, its CI and the two-window model RMSE are not estimable; reported as null and conclusion set to premise_invalid. No proxy or substitute window was used."
    },
    {
      "issue": "The 120-trading-day pre-window is defined relative to the first flagged day, which does not exist in the file.",
      "handling": "Pre-window not constructed; not substituted by the last 120 rows of 2019."
    },
    {
      "issue": "Calendar gaps between consecutive rows up to 5 days (6 gaps >4 days), consistent with weekends and Canadian holidays; no missing-value rows.",
      "handling": "One-day changes computed between consecutive rows as specified by TASK.md, regardless of calendar gap."
    },
    {
      "issue": "First row has no previous observation, so no one-day change.",
      "handling": "Excluded from change-based statistics (2497 changes)."
    }
  ],
  "n_input_rows": {
    "gc5y_daily.csv": 2498
  },
  "n_analysed": null,
  "random_seed": null
}