# Overnight Research Loop

> Runs overnight crawling of news/publications to generate factor hypotheses, runs backtests in parallel, debates findings, and generates a morning briefing.

- Skill: `knuckles-team/overnight-research-loop` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add knuckles-team/overnight-research-loop`
- Raw SKILL.md: https://api.skillmd.com/api/skills/knuckles-team/overnight-research-loop/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: Knuckles-Team (https://skillmd.com/u/knuckles-team)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/knuckles-team/overnight-research-loop

---


# Overnight Research Loop Workflow

**CONCEPT:EE-011**

Runs overnight crawling of news/publications to generate factor hypotheses, runs backtests in parallel, debates findings, and generates a morning briefing.

## Steps

### Step 1: Hypothesis Crawlers
**Agent**: `data-fetcher`
**Tools**: `graph_query, sx_search`

Scans academic publications, social feeds, and market tickers overnight to find potential factor ideas.
Expected: `factor-hypothesis-list`

### Step 2: Parallel Backtester [depends_on: hypothesis-crawlers]
**Agent**: `compute-engine`
**Tools**: `graph_analyze`

Runs backtests concurrently across 50 parameter configurations.
Expected: `multi-config-backtest-results`

### Step 3: Factor Debater [depends_on: parallel-backtester]
**Agent**: `risk-assessor`
**Tools**: `graph_query, graph_analyze`

Analyzes return profile, drawdown, and transaction cost impact in a swarm debate.
Expected: `approved-factor-signals`

### Step 4: Report Compiler [depends_on: factor-debater]
**Agent**: `report-generator`
**Tools**: `graph_write, document_tools`

Compiles findings into a formatted dashboard report for the morning review.
Expected: `morning-research-tearsheet`

### Step 5: KG Persistence [depends_on: report-compiler]
**Agent**: `report-generator`
**Tools**: `graph_write`

Persist workflow results as nodes and edges in the Knowledge Graph.
Create appropriate typed nodes with metadata and link to existing domain entities.

## Output
- Overnight Research Loop results persisted in KG
- Structured report (MD/PDF)
- Audit trail with timestamps and agent attributions

## Execution

Run this workflow as a dependency-ordered DAG. Steps with no unmet `depends_on` run in parallel; dependents run after their prerequisites complete.

- **Run first (in parallel):** Step 1 — Hypothesis Crawlers
- **After level 0:** Step 2 — Parallel Backtester
- **After level 1:** Step 3 — Factor Debater
- **After level 2:** Step 4 — Report Compiler
- **After level 3:** Step 5 — KG Persistence

**Execution:** If graph-os is reachable, offload the whole DAG via `graph_orchestrate action=execute_workflow` (or the `kg-delegate` skill) for true parallel/swarm execution. Otherwise execute the steps natively in dependency order: run steps with no unmet `depends_on` in parallel, then their dependents.