rgourley historical-comparison
Twin comparison: event-study on the ticker + historical-analog-finder on the market. The idea: don't rely on one anchor when you can use both.
When to invoke
- Analyst wants both "here's what happened around this specific event" and "here's what usually happens in setups like this"
- Before a directional call where both name-specific and market context matter
- Analog-only mode (no ticker): just the market regime forward distribution
Modes
- Event mode: pass ticker + event_class + optional event_date / period. Both event and analog run.
- Analog-only: pass
include_event=False. Only the market analog runs.
What you get back
Layer 1 JSON matching output-schema.json.
Headline distills the T+5 CAR + prior percentile from event-study
plus the median/IQR/hit-rate at 90d from the analog.
Layer 2 rendered brief. See
references/rendering.md.