Skip to content

Quant trading agent: Get minute-level, MBB-grade investment insights powered by SEC and 10-K filings.

License

Notifications You must be signed in to change notification settings

raucvr/Group-Goki

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Group-Goki — Quant Trading Agent

Get minute-level, MBB-grade investment insights powered by SEC and 10-K filings.

MIT License

Group-Goki is an AI-powered quantitative trading agent that monitors SEC EDGAR for new filings, automatically escalates polling frequency as filing deadlines approach, and delivers structured investment analysis reports using a 6-section MECE framework modeled on top-tier consulting methodology.

  • Real-time SEC EDGAR monitoring — watches for 10-K, 10-Q, and 8-K filings with adaptive polling (5min → 1min as deadlines approach)
  • Automated 10-K parsing — extracts Business, Risk Factors, MD&A, and Financial Statements sections with revenue/EPS highlights
  • MBB-grade analysis framework — bull/bear/neutral thesis with conviction levels, structured like a McKinsey/BCG deliverable
  • Zero manual intervention — lock a filing date, and the system handles escalation, detection, analysis, and delivery autonomously

How it works

                         ┌─────────────────────────┐
                         │       SEC EDGAR          │
                         │  (EFTS + Submissions)    │
                         └────────────┬─────────────┘
                                      │
                                      ▼
┌──────────────┐          ┌───────────────────────┐
│  Watchlist    │◄────────│   SEC EDGAR Client     │
│  (JSON store) │         │  (rate-limited fetch)  │
└──────┬───────┘          └───────────────────────┘
       │
       ▼
┌──────────────────────────────────────────────────┐
│          Filing Monitor Service (background)      │
│                                                   │
│  Escalation Ladder:                               │
│    > 24h  → no cron (watching)                    │
│    2-24h  → poll every 5 min (pre-filing)         │
│    < 2h   → poll every 1 min (imminent)           │
│                                                   │
│  On new filing detected:                          │
│    → fetch + parse 10-K sections                  │
│    → generate 6-section MECE analysis             │
│    → deliver to configured channel                │
│    → deescalate monitoring                        │
└──────────────────────────────────────────────────┘

Tools & Actions

Group-Goki exposes three tools with 14 actions total:

sec_filing — SEC EDGAR data access

Action Description
search Full-text search across EDGAR filings by ticker, form type, and date range
company Fetch company profile and recent submissions (auto-resolves ticker → CIK)
filing Retrieve raw filing document by URL or accession number
sections Parse 10-K into key sections with financial highlights extraction

filing_schedule — Watchlist management

Action Description
list Show all watched companies with time-until-filing countdown
add Add a company (auto-resolves ticker → CIK via SEC)
remove Remove a company from the watchlist
update Update expected filing date/time
lock Lock the filing window — triggers automated monitoring escalation
status Get detailed status for a specific ticker

heartbeat_control — Cron job management

Action Description
activate Start periodic SEC EDGAR polling for a ticker
escalate Shorten the polling interval (e.g., 5min → 1min)
deescalate Stop monitoring and clean up cron jobs
status List all active monitoring jobs

10-K Analysis Framework

When a new filing is detected, the agent produces a structured report with six sections:

  1. Executive Summary — Bull/bear/neutral thesis with conviction level (1-5)
  2. Business Quality Assessment — Revenue growth, margins, competitive moat analysis
  3. Risk Factor Analysis — New/removed risks compared to prior filing, severity ranking
  4. MD&A Deep Dive — Guidance vs. actuals, forward-looking statements, capital allocation
  5. Financial Statement Highlights — Revenue, EPS, balance sheet strength, free cash flow
  6. Investment Thesis — Bull/bear/base cases with probability-weighted expected returns

Quick Start

Prerequisites: Node.js >= 22, pnpm

git clone https://github.com/raucvr/Group-Goki.git
cd Group-Goki

pnpm install
pnpm build

Configure your SEC EDGAR User-Agent (required by SEC):

// ~/.groupgoki/groupgoki.json
{
  plugins: {
    "quant-sec": {
      userAgent: "YourCompany you@example.com",
    },
  },
}

Or set the SEC_USER_AGENT environment variable:

export SEC_USER_AGENT="YourCompany you@example.com"

Configuration

Key Type Required Default Description
userAgent string Yes SEC EDGAR User-Agent (format: "CompanyName admin@company.com")
rateLimitRps number No 8 Max requests/sec to SEC EDGAR (hard cap: 10)
deliveryChannel string No Channel for delivering analysis reports (e.g., "telegram")
deliveryTo string No Recipient identifier for the delivery channel

Project Structure

extensions/quant-sec/
├── index.ts                    # Plugin entry — registers tools + service
└── src/
    ├── sec-edgar-client.ts     # HTTP client for SEC EDGAR APIs (rate-limited)
    ├── sec-filing-parser.ts     # 10-K section parser + financial highlights
    ├── sec-tool.ts             # sec_filing tool (search/company/filing/sections)
    ├── filing-schedule-tool.ts # filing_schedule tool (list/add/remove/update/lock/status)
    ├── heartbeat-control-tool.ts # heartbeat_control tool (activate/escalate/deescalate/status)
    ├── filing-monitor-service.ts # Background auto-escalation service
    ├── store.ts                # Immutable JSON watchlist persistence
    ├── config.ts               # Plugin config validation
    ├── types.ts                # TypeScript type definitions
    ├── utils.ts                # Shared helpers (cron naming, state dir)
    └── __tests__/              # 108 tests across 9 test files

Testing

# Run all quant-sec tests
pnpm test extensions/quant-sec

# Verbose output
pnpm test extensions/quant-sec -- --reporter=verbose

108 tests across 9 test files covering:

  • SEC EDGAR client (fetch mocking, URL building, rate limiter, CIK resolution)
  • All 3 tools and 14 actions (with full mock isolation)
  • 10-K parser (section extraction, financial highlights)
  • Watchlist store (immutable CRUD operations)
  • Auto-escalation state machine
  • Config validation

Built on OpenClaw

Group-Goki is built as an extension on OpenClaw, an open-source personal AI assistant platform. OpenClaw provides the gateway, agent runtime, cron scheduler, multi-channel delivery, and plugin system that Group-Goki uses for its monitoring and delivery pipeline.

License

MIT — see LICENSE for details.

About

Quant trading agent: Get minute-level, MBB-grade investment insights powered by SEC and 10-K filings.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Sponsor this project

Packages

 
 
 

Contributors