{"empty":false,"at":"2026-11-13T14:35:17.652Z","params":{"version":1,"recal_a":1,"recal_b":0,"blend_w":0.5,"last_tune_at":null,"domains_off":{}},"settled":3,"forecasts":137,"tune_after":50,"config":[{"key":"max_updates","value":"5","updated_at":"2026-11-12T15:15:24.372Z","reason":"coverage 1, 810 tokens a forecast and 12 open questions: one more update an hour"},{"key":"broadcast_format","value":"registered-first","updated_at":"2026-11-12T15:15:24.372Z","reason":"experiment period: arm registered-first"}],"experiments":[{"key":"broadcast_format","arm":"registered-first","started_at":"2026-11-12T15:15:24.372Z","ended_at":null,"score":null}],"research":[],"research_seen":0,"events":[{"at":"2026-11-12T23:35:26.434Z","kind":"parameter","what":"pause new signal entries until 2026-11-13T23:35:26.434Z","why":"drawdown 19.7% reached the 15% breaker","revert":"DELETE FROM pt_control WHERE key IN ('paused_until','paused_reason')"},{"at":"2026-11-12T15:15:24.372Z","kind":"knob","what":"max_updates: 4 → 5","why":"coverage 1, 810 tokens a forecast and 12 open questions: one more update an hour","revert":"DELETE FROM fe_config WHERE key = 'max_updates'"},{"at":"2026-11-12T15:15:24.372Z","kind":"knob","what":"undefined: undefined → undefined","why":"","revert":""},{"at":"2026-11-12T15:15:24.372Z","kind":"experiment running","what":"broadcast_format: arm registered-first","why":"sample still collecting","revert":""},{"at":"2026-11-11T23:35:24.373Z","kind":"parameter","what":"pause new signal entries until 2026-11-12T23:35:24.373Z","why":"drawdown 15.6% reached the 15% breaker","revert":"DELETE FROM pt_control WHERE key IN ('paused_until','paused_reason')"}],"loop":[["Measure","Every forecast is hash-chained before its event and scored against the market price when it settles (Brier score). That score is the target the system optimises."],["Propose","Once enough calls have settled, the tuner fits a recalibration and a model-versus-market blend; the daily product manager proposes a change to a knob; the research lane turns a published method into a code task."],["Trial","A change runs as one experiment at a time with its own sample. Calibration changes start a new parameter version, so old and new are never mixed."],["Keep or revert","The result decides: the winner is written to the config, the loser is reverted, a domain where the model is significantly worse than the market is paused, and every step keeps its reason and its revert statement."],["Rewrite","A method that needs new code becomes a code task with a test; the code loop builds it, the deploy probes it, and the next measurement says whether it helped."]]}