# Contract (P0) — F6-conf-01

Question: Does the weekday carry predictive content for the daily change in the 10y Treasury CMT yield (DGS10), 2000–2019? Estimands:

1. Joint test of equal weekday means of Δy_t (bp) — classical one-way equal-variance ANOVA F, F(4, n−5), all changes.
2. Monday contrast = mean(Δy | Monday) − mean(Δy | Tue–Fri), all changes, with 95% CI. Choice: Welch two-sample t interval (unequal variances, Welch–Satterthwaite df).
3. Out-of-sample: RMSE over evaluation period (changes dated >= 2010-01-01) of zero-change forecast (baseline) vs. weekday-specific means estimated on changes dated <= 2009-12-31. Assumptions: changes treated as independent draws within weekday (ANOVA additionally assumes equal variances, approx. normality — large n mitigates); stationarity of weekday means across periods for OOS. Information rules: drop empty DGS10 rows first; Δy_t = 100(y_t − y_{t−1}) between consecutive available obs, attributed to weekday of t; first obs has no change. Weekday means for forecasting from estimation period only. Only DGS10.csv used. Outputs: out/results.json (n_missing_rows, n_train, n_test, anova_f_stat, anova_p_value, effect_estimate, effect_ci_low, effect_ci_high, baseline_metric, model_metric, conclusion, task_id, feasibility, feasibility_reason, data_issues, n_input_rows, n_analysed, random_seed); out/figure1.png; out/table1.csv (weekday,n,mean_bp,se_bp); out/report.md (6 sections); out/analysis.py; out/claims.json. Conclusion rule: ANOVA p>=0.05 → no_detectable_effect; if rejects, sign from Monday CI (positive/negative; if CI covers zero despite ANOVA rejection → no_detectable_effect for the Monday-contrast estimand, documented). Invalidating conditions: missing/corrupt file, hash mismatch, no data in either period, weekdays missing, use of evaluation-period info in forecasts.