# Contract (P0) — F3-conf-02
**Question.** How accurately could US quarterly real GDP growth (A191RL1Q225SBEA, % q/q SAAR) have been forecast 1 and 2 quarters ahead in real time, using only the vintage published at each origin, and does a simple time-series model beat the naive (last value) forecast?

**Estimands.** Over 8 origins (vintage dates 2024-01-31 … 2025-12-31) and h∈{1,2} quarters past the last quarter in the origin vintage: RMSE and MAE of (a) naive and (b) one model, vs first-release values of the targets; relative improvement 1−RMSE_model/RMSE_naive per h; mean |current − first release| over the 8 h=1 target quarters. Meaningful under: first release as the evaluation "truth"; strict real-time use of vintages; with only 8 errors per horizon, estimates are very noisy (no inferential claim of superiority is warranted).

**Outputs.** out/forecasts.csv (32 rows; origin_date,target_date,horizon,forecast,actual,model,vintage_date,train_last_date,train_last_value,n_train); out/results.json (task_id, model_name, model_description, n_origins, first_origin, last_origin, rmse/mae naive & model h1/h2, rel_improvement_rmse_h1/h2, mean_abs_revision_h1_targets, feasibility, feasibility_reason, data_issues, n_input_rows, n_analysed, random_seed); out/report.md (Question; Data and exclusions; Method; Results; Uncertainty; Limitations and open issues); out/analysis.py; out/claims.json.

**Information rules.** At origin d, only vintage file dated d is used for estimation/forecast input. current_vintage.csv only for the revision statistic. Actuals = first_release.csv. Naive = last value in origin vintage for both horizons.

**Invalidity/impossibility.** Missing vintage or first-release value for any target; using current vintage or later vintages in estimation; first-release vintage date of a target earlier than or equal to origin (would mean target was already known).