edge-hint-extractor

---
name: edge-hint-extractor
description: Use when converting daily market observations, anomalies, and news reactions into structured hints.yaml for downstream concept synthesis or auto-detection.
category: Strategy R&D
version: 1.0.0
tools: []
---

# Edge Hint Extractor

## Overview

Convert raw observation signals (`market_summary`, `anomalies`, news reactions) into structured edge hints. This is the first stage in the split workflow: observe -> abstract -> design -> pipeline.

## When to Use

- Turn daily market observations into reusable hint objects.
- Generate LLM ideas constrained by the current anomalies/news context.
- Produce a clean `hints.yaml` input for concept synthesis or auto-detection.

## Inputs

- `market_summary.json`, `anomalies.json`, and optionally `news_reactions` (CSV or JSON) from a detector run.

## Output

`hints.yaml` containing the hint list plus generation metadata and rule/LLM hint counts.

## Methodology

1. Gather observation files: market summary, anomalies, and optional news reactions.
2. Generate deterministic, rule-based hints directly from these observations.
3. Optionally augment with LLM-generated ideas in one of two ways: (a) an external LLM pass constrained to the current anomalies/news context, or (b) hints already drafted by the calling agent itself (e.g. Claude reasoning over the anomalies directly and writing hint objects) — these two augmentation paths are mutually exclusive, pick one.
4. Each hint should carry: a title, the source observation, relevant symbols, regime bias, a mechanism tag, a preferred entry family, and a hypothesis type.
5. Pass the resulting `hints.yaml` into concept synthesis or auto-detection as the next stage.