| Name | Target | Horizon | Folds | RMSE | Coverage 80 | Trained |
|---|---|---|---|---|---|---|
| crack_321_change_7d_lgbm_v1783883330 crack 3-2-1 + regime features (inv/util percentiles, tight flags) [suggestion:crack-321-target] | crack_321_change_7d | 7d | 4 | 3.9579 | 66% | 2026-07-12 19:08 |
| wti_return_7d_chronos_bolt_v1783882365 Zero-shot Chronos-Bolt on the baseline's exact folds — univariate price history only [suggestion:foundation-model-benchmark] | wti_log_return_7d | 7d | 4 | 0.0681 | 73% | 2026-07-12 18:52 |
| wpsr_dist_direction_v1783882248 Draw/build direction classifier for the Wednesday WPSR print [suggestion:dist-stocks-direction] | distillate_stocks_direction_1w | 7d | 6 | — | — | 2026-07-12 18:50 |
| wti_return_7d_lgbm_v1783882150 COT managed-money net as feature [suggestion:cot-coiled-spring] | wti_log_return_7d | 7d | 4 | 0.0662 | 69% | 2026-07-12 18:49 |
| crack_321_change_7d_lgbm_v1783882122 Gulf 3-2-1 crack, 7d change — first product-thesis model; conformal + lr0.01 [suggestion:crack-321-target] | crack_321_change_7d | 7d | 4 | 3.9675 | 68% | 2026-07-12 18:48 |
| wti_return_7d_lgbm_v1783881807 conformal-calibrated bands [suggestion:conformal] | wti_log_return_7d | 7d | 4 | 0.0678 | 60% | 2026-07-12 18:43 |
| wti_return_7d_lgbm_v1783881776 macro-only: no oil prices (TSA excluded — series starts 2026-01) [suggestion:macro-only] | wti_log_return_7d | 7d | 3 | 0.0787 | 35% | 2026-07-12 18:42 |
| wti_return_7d_lgbm_v1783881764 p25/p75 band — calibration across quantile levels [suggestion:narrower-bands] | wti_log_return_7d | 7d | 4 | 0.0713 | 61% | 2026-07-12 18:42 |
| wti_price_7d_lgbm_v1783881715 absolute price target — stationarity lesson [suggestion:predict-price] | wti_price_7d | 7d | 4 | 5.6943 | 54% | 2026-07-12 18:41 |
| wti_return_7d_lgbm_v1783881683 LR sweep high [suggestion:vary-learning-rate] | wti_log_return_7d | 7d | 4 | 0.0724 | 51% | 2026-07-12 18:41 |
| wti_return_7d_lgbm_v1783881652 LR sweep low [suggestion:vary-learning-rate] | wti_log_return_7d | 7d | 4 | 0.0657 | 70% | 2026-07-12 18:40 |
| wti_return_7d_lgbm_v1783881619 ablation: no WTI features [suggestion:ablation-wti-lags] | wti_log_return_7d | 7d | 4 | 0.0699 | 62% | 2026-07-12 18:40 |
| wti_return_1d_lgbm_v1783881588 1d horizon — daily noise test [suggestion:shorter-horizon] | wti_log_return_1d | 1d | 4 | 0.0257 | 70% | 2026-07-12 18:39 |
| wti_return_30d_lgbm_v1783881556 30d horizon — uncertainty decay test [suggestion:longer-horizon] | wti_log_return_30d | 30d | 4 | 0.1260 | 53% | 2026-07-12 18:39 |
| wti_return_7d_lgbm_v2_enriched v2: defaults + demand proxies (CPC degree days), Henry Hub, retail prices (margin signal), 10y yield, refinery capacity. 1997+-compatible set to preserve training span. Trained after Tier 0-2 data platform build. | wti_log_return_7d | 7d | 4 | 0.0698 | 56% | 2026-07-11 20:23 |
| wti_return_7d_lgbm_v1783799602 Walk-forward LightGBM quantile (p10/p50/p90). | wti_log_return_7d | 7d | 4 | 0.0713 | 61% | 2026-07-11 19:53 |
| wti_return_7d_lgbm_v1777334304 real run after fold-overlap fix | wti_log_return_7d | 7d | 4 | 0.0627 | 62% | 2026-04-27 23:58 |
| wti_return_7d_lgbm_v1777329440 first real run — 36 years of EIA + FRED data, 11 feature series × 3 lags | wti_log_return_7d | 7d | 2 | 0.0743 | 61% | 2026-04-27 22:37 |
| wti_return_7d_lgbm_v1777221296 smoke test on Q1 2024 only — pipeline verification, not a real model | wti_log_return_7d | 7d | 1 | 0.0271 | 93% | 2026-04-26 11:34 |