{ "schema_version": 1, "id": "ext_checkpoint_forecast_validation", "question_id": "r08_early_blowup_signal", "title": "How early can splits reveal the finish?", "status": "ready", "as_of": "2026-09-12T14:13:07Z", "n": 869761, "input_export_id": "private-20260912-0934", "input_as_of": "2026-09-12T13:37:16Z", "live_json_as_of": "2026-09-12T04:11:06Z", "input_asset_sha256": "cb4e0ab7432010de29755ec5d75ca3b370e64e481eda5b20c44b359f39552204", "input_manifest_sha256": "a6f534d40bac291e0a1bc2eeb70633660acd00a710a090320876736de9c30f4d", "analysis_script_sha256": "4a26017aa2c2b2796e8d5f3498ea5e0a2f3c1e691a7332ef20cdbc256d4cdd42", "analysis_version": 2, "engine": "DuckDB 1.4.4", "corpus": { "n_records": 4462379, "n_cities": 34, "n_race_years": 256 }, "cohort": { "deduplicated": 4462379, "complete": 3830211, "increasing": 3604993, "timing_eligible": 3595426, "eligible": 3517336, "source_quality_excluded": 78090, "raw": 4462379, "duplicates_removed": 0, "missing_or_unparsed": 632168, "non_increasing": 225218, "outside_quality_bounds": 9567 }, "methodology_prose": [ "At each checkpoint, start with elapsed time \u00d7 42.195 / distance. The elapsed-only model multiplies this by the training median actual/projected ratio in 30-second-per-km elapsed-pace bands. The trend model adds the most recent section\u2019s change from the previous section (faster than \u22122%, within \u00b12%, or slower than +2%). At 5 km no trend exists.", "Hold out the latest three observed calendar years (2024\u20132026). Fit every factor and the 10th\u201390th percentile ratio interval on earlier years only. Training cells need at least 100 records; missing trend cells fall back to the pace-band model, then the pooled training model. No future splits, finishing-time groups, identities or supplied ability fields enter a prediction.", "Report median absolute error for all three methods, plus observed coverage and median width of the nominal 80% prediction interval and the 90th percentile absolute error. Model selection is fixed before examining these results. A runner may occur in training and test in different years; identities are not used. Results apply to complete eligible finishers and do not predict withdrawals.", "Use complete, strictly increasing elapsed checkpoints at 5, 10, 15, 20, 25, 30, 35, 40 and 42.195 km. Clock strings must parse as H:MM:SS or M:SS. No missing splits are interpolated.", "Remove exact duplicate race records, ignoring database IDs, ingestion timestamps and source URLs. Retain finishes from 90 minutes to 12 hours with every section between 2 and 20 minutes per km. These quality filters can exclude genuine unusual performances; the analysis describes this eligible cohort, not every entrant.", "Full source records are available in public GitHub Releases. These chart tables require at least 100 eligible observations per cell for estimate reliability. Counts refer to finishes, linked pairs or event observations as specified in that answer.", "These are observational results. Fitness changes, intentions, training, selection into the dataset and unmeasured conditions can explain differences. Outcome percentiles describe variation among performances, not confidence intervals or advice about the best strategy.", "Apply the reviewed source-quality edition exclusions for this exact export after the timing checks. Known invalid split grids, incomplete ingestion, unreconciled HOLD editions and a selected top-finisher field do not contribute to the analyses or prior benchmarks. Report source exclusions separately; an already invalid timing row is not counted twice. Missing age or recorded gender alone does not exclude an otherwise eligible finish from the overall cohort. Other sparse editions are not declared incomplete merely from their size." ], "observational": true, "minimum_public_cell": 100, "source_quality": { "version": 1, "release_tag": "private-export-20260912-0934", "reviewed_edition_policy": true, "editions": [ { "city": "Honolulu", "year": 2016, "category": "invalid_split_grid", "reason": "Producer explicitly force-dropped this incomplete 5 km grid.", "raw_records": 20118, "deduplicated_records": 20118, "timing_eligible_excluded": 0 }, { "city": "Rotterdam", "year": 2012, "category": "invalid_split_grid", "reason": "Producer explicitly force-dropped this incomplete 5 km grid.", "raw_records": 7542, "deduplicated_records": 7542, "timing_eligible_excluded": 0 }, { "city": "Eindhoven", "year": 2009, "category": "invalid_split_grid", "reason": "Producer explicitly force-dropped this incomplete 5 km grid.", "raw_records": 1407, "deduplicated_records": 1407, "timing_eligible_excluded": 0 }, { "city": "Eindhoven", "year": 2010, "category": "invalid_split_grid", "reason": "Producer explicitly force-dropped this incomplete 5 km grid.", "raw_records": 1462, "deduplicated_records": 1462, "timing_eligible_excluded": 0 }, { "city": "Frankfurt", "year": 2019, "category": "invalid_split_grid", "reason": "Producer explicitly force-dropped this 5 km grid; numerically plausible rows are not reinstated.", "raw_records": 12513, "deduplicated_records": 12513, "timing_eligible_excluded": 10624 }, { "city": "New York", "year": 2008, "category": "unreconciled_hold", "reason": "Previously partial ingestion has grown to 38,047 records, but source-total reconciliation and explicit completion evidence are absent; keep the prior hold pending review.", "raw_records": 38047, "deduplicated_records": 38047, "timing_eligible_excluded": 37387 }, { "city": "Chicago", "year": 2016, "category": "incomplete_ingestion", "reason": "Producer handoff identifies this 6,512-record edition as incomplete and on HOLD.", "raw_records": 6512, "deduplicated_records": 6512, "timing_eligible_excluded": 6506 }, { "city": "Tokyo", "year": 2026, "category": "selected_field", "reason": "Source notes identify top-500 PDFs, not a representative full marathon field.", "raw_records": 353, "deduplicated_records": 353, "timing_eligible_excluded": 353 }, { "city": "Amsterdam", "year": 2025, "category": "unreconciled_hold", "reason": "Source audit marks HOLD: 23,328 listed versus 23,325 retrieved. This conservative hold does not establish a large bias from the three missing records.", "raw_records": 23325, "deduplicated_records": 23325, "timing_eligible_excluded": 23220 }, { "city": "Valencia", "year": 2017, "category": "incomplete_ingestion", "reason": "Source audit marks HOLD: 3,373 listed detail pages missing; the observed records also lack 20 km.", "raw_records": 12485, "deduplicated_records": 12485, "timing_eligible_excluded": 0 }, { "city": "Valencia", "year": 2018, "category": "incomplete_ingestion", "reason": "Producer handoff retains an incomplete/HOLD status; the edition also lacks the observed 20 km checkpoint.", "raw_records": 19236, "deduplicated_records": 19236, "timing_eligible_excluded": 0 } ], "policy_sha256": "c08e83c993025dc41c0344be7b5442d74b3fb0588c38949481d5bc0a08f45fa0", "script_sha256": "433e8b9566b5afd809f4de1180fab222e63ad40678a33f6c93f08712cf820c77" }, "source_quality_script_sha256": "433e8b9566b5afd809f4de1180fab222e63ad40678a33f6c93f08712cf820c77", "observation_unit": "eligible finishes", "evidence_scope": "validated forecast", "supporting_script_sha256": "7c6959b427fa1b06bb9600bb3377ecb226257d14aa0ec61a71a63f42bc87a3dd", "linkage_audit": { "feature_rows": 4462379, "safe_identity_groups": 2892236, "record_join_method": "canonical_record_id", "canonical_id_contract_verified": true, "candidate_link_rows": 3194070, "natural_key_candidate_rows": 0, "canonical_id_candidate_rows": 3194070, "ambiguous_cross_export_matches": 0, "linked_eligible_finishes": 3194070, "recent_benchmark_finishes": 555437, "consecutive_cross_year_pairs": 583670, "linked_with_supplied_date": 3185321 } }