{"data":{"slug":"calibration-curve","term":"Calibration Curve","bucket":"foresight","definition":"A chart of predicted probability against actual frequency. A well-calibrated model sits on the diagonal of the curve.","short_definition":"A chart of predicted probability against actual frequency. A well-calibrated model sits on the diagonal of the curve.","long_definition":"Calibration Curve buckets Foresight predictions by confidence and plots the actual outcome rate inside each bucket. A model that says 70% should land near 70% reality. Foresight publishes the curve alongside every backtest so a reader can audit over-confidence and under-confidence directly, without re-running the model.","related_terms":["foresight","backtest-run","brier-score"],"related_term_routes":[{"slug":"foresight","term":"Foresight","route":"/resources/glossary/foresight"},{"slug":"backtest-run","term":"Backtest Run","route":"/resources/glossary/backtest-run"},{"slug":"brier-score","term":"Brier Score","route":"/resources/glossary/brier-score"}],"canonical_route":"/resources/glossary/calibration-curve","api_route":"/api/v1/resources/glossary/calibration-curve","jsonld_id":"https://sciencetostartup.com/resources/glossary/calibration-curve","variants":[],"tldr":"A chart of predicted probability against actual frequency. A well-calibrated model sits on the diagonal of the curve.","key_points":[],"quality_tier":null,"citation_count":null,"source_state":"curated_static","source_module":"apps/web/data/glossary/terms.ts","definition_sections":{"schema_version":1,"intro":"Calibration Curve buckets Foresight predictions by confidence and plots the actual outcome rate inside each bucket. A model that says 70% should land near 70% reality. Foresight publishes the curve alongside every backtest so a reader can audit over-confidence and under-confidence directly, without re-running the model.","sections":[{"title":"Definition","items":[{"subtitle":"Calibration Curve","text":"Calibration Curve buckets Foresight predictions by confidence and plots the actual outcome rate inside each bucket. A model that says 70% should land near 70% reality. Foresight publishes the curve alongside every backtest so a reader can audit over-confidence and under-confidence directly, without re-running the model."}]},{"title":"Related vocabulary","items":[{"subtitle":"Foresight","text":"The verifiable-prediction surface. Public ledger of frozen predictions with backtests, reasoning chain, and a self-improving flywheel."},{"subtitle":"Backtest Run","text":"A historical evaluation of the scoring model against known outcomes. Returns hit rate, precision@k, calibration curve."},{"subtitle":"Brier Score","text":"A proper scoring rule for probabilistic predictions. Lower is better. Public Brier receipt is a pending gate for Wave 7."}]}],"cited_arxiv_ids":[]}},"meta":{"canonical_route":"/resources/glossary/calibration-curve","api_route":"/api/v1/resources/glossary/calibration-curve","source":{"label":"curated glossary catalog","source_state":"curated_static","source_module":"apps/web/data/glossary/terms.ts","method_version":"public_glossary_curated_terms_v2","freshness":{"status":"versioned","observed_at":null,"fresh_until":null,"reason":"Git-versioned curated catalog; daily ingestion freshness windows do not apply.","reason_code":"git_versioned_curated_catalog"},"source_count":111,"bucket_count":7,"buckets":["scoring","surfaces","agents","distribution","data","foresight","buildability"]}}}