# Contract — F6-conf-01

Question. Does the weekday predict the one-day change in the 10y Treasury CMT yield (DGS10), 2000–2019? Estimands. (i) ANOVA (classical equal-variance one-way) F-test of equal weekday means of Δy_t (bp) on all changes; (ii) Monday contrast = mean(Mon) − mean(Tue–Fri), with 95% CI (choice: Welch two-sample t interval); (iii) out-of-sample RMSE (2010–2019 changes) of weekday-mean forecast (means estimated on changes dated ≤2009-12-31) vs zero-change forecast. Assumptions: changes are roughly independent across days for the ANOVA/Welch inference to be valid (i.i.d. within groups; ANOVA also assumes equal variance); stationarity of weekday means across periods for OOS meaning. Definitions. Drop empty DGS10 rows first; Δy_t = 100(y_t − y_{t−1}) between consecutive available observations; dated/attributed to weekday of t; first obs has no change. Train = changes dated ≤2009-12-31 (includes the first change of 2000; the change dated 2010-01-04 spanning year boundary belongs to test). Outputs. out/results.json (keys: 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 by Monday CI (positive/negative; if CI covers zero though ANOVA rejects → no_detectable_effect for the Monday sign, documented). Info rules. Only data/DGS10.csv; weekday means for forecasting from train period only. Invalidating conditions. Missing/corrupt file, inability to compute changes; using test data in forecast means.