# MODELS.md
> Edit Guidance
> - Keep entries append-only by date. Do not delete prior model notes.
> - Use concrete evidence and outcomes; avoid vague summaries.
> - If updating an existing model, add a new dated revision block instead of overwriting history.
> - Record both wins and failures with explicit lesson statements.
## Purpose
Track reusable cause-effect models, calibration stats, and lessons from scored predictions.
## Entry Template
### YYYY-MM-DD — Model Name
- Timeframe: `low|medium|high|macro`
- Claim: `[cause] -> [effect]`
- Inputs watched:
- Validation signals:
- Invalidation signals:
- Outcome:
- Confidence adjustment:
- Lesson:
## Active Models
### (add new model blocks above this line)