OpenQuant

The open-source operating system for quant trading

Works without any API key. Zero config. Zero cost.

---

This is a proof of concept for an AI-powered personal hedge fund. It uses a team of AI agents to analyze stocks, manage risk, and execute strategies — from your terminal.

---

The Agents

Agent	What It Does
🕵️ Insider Sentiment	Tracks Form 4 filings. Scores insider buying/selling momentum.
📈 Momentum	Detects breakouts, trend shifts, and regime changes from OHLCV data.
🔍 Value Deep	Digs into fundamentals — P/E, FCF, margins, balance sheet health.
💰 Earnings Surge	Monitors EPS surprises, guidance shifts, and post-earnings drift.
📊 Analyst Consensus	Aggregates Wall Street estimates and tracks revision momentum.

Each agent generates independent signals. A risk engine sizes positions. You decide what to trade.

---

30-Second Demo

# Install
pip install openquant-cli

# Run — no API key needed (uses yfinance by default)
openquant analyze AAPL

# Get structured JSON for your own agent
openquant --json analyze AAPL

# Check risk
openquant risk TSLA

# Track insiders
openquant insider NVDA

# Run a strategy
openquant strategy run insider-momentum -t AAPL

$ openquant analyze AAPL

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  AAPL · Apple Inc. · Technology · Consumer Electronics
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

  Price:    $198.42    Vol: 52,841,300    MCap: $3.01T
  52w High: $260.10    52w Low: $164.08

  ┌─ Signals ──────────────────────────────────────────┐
  │  🕵️ Insider Sentiment    BULLISH  (+0.72)          │
  │     3 insider buys in last 30 days                  │
  │                                                     │
  │  📈 Momentum            NEUTRAL  (+0.18)           │
  │     Price below 50-day MA, RSI: 44.2               │
  │                                                     │
  │  🔍 Value Deep           BULLISH  (+0.65)          │
  │     P/E 28.4x, FCF yield 3.8%, margins expanding   │
  │                                                     │
  │  💰 Earnings Surge       BULLISH  (+0.81)          │
  │     Last EPS: +8.2% surprise, guidance raised       │
  │                                                     │
  │  📊 Analyst Consensus    BULLISH  (+0.54)          │
  │     22 Buy / 6 Hold / 2 Sell, PT $225 avg          │
  └─────────────────────────────────────────────────────┘

  Risk:    Vol 22.4%  |  MaxDD 13.8%  |  Sharpe 0.99  |  VaR95 -2.0%

---

How It Works

  Your Agent (Claude Code, Hermes, Codex...)
       │
       ▼
  ┌─────────────────────────────────────────┐
  │           OpenQuant Engine              │
  │                                         │
  │  ┌──────────┐  ┌──────────┐  ┌───────┐ │
  │  │ Data      │  │ Compute  │  │ Exec  │ │
  │  │           │  │          │  │       │ │
  │  │ yfinance  │  │ 5 agents │  │ Paper │ │
  │  │ SEC EDGAR │  │ Risk     │  │ Alpaca│ │
  │  │ QuantFetch│  │ Strategy │  │ Kalshi│ │
  │  └──────────┘  └──────────┘  └───────┘ │
  │                                         │
  │  Interfaces: CLI · TUI · MCP · JSON    │
  └─────────────────────────────────────────┘
       │              │            │
       ▼              ▼            ▼
    Terminal      AI Agents    Your Code

Three layers, one engine:

1. Data — yfinance (free), SEC EDGAR (free), or QuantFetch API (premium)
2. Compute — 5 analysis agents, risk engine, Monte Carlo VaR, strategy framework
3. Execution — Paper trading, Alpaca (stocks), Kalshi (prediction markets)

---

Built-in Strategies

Strategy	Signal	Style
insider-momentum	Insider buying + price momentum	Aggressive growth
value-deep	Fundamental value + margin expansion	Patient value
earnings-surge	EPS surprise + guidance drift	Event-driven
technical-breakout	Breakout + volume confirmation	Momentum

# List all strategies
openquant strategy list

# Run a strategy in paper mode
openquant strategy run insider-momentum -t AAPL -m paper

# Run in game mode (gamified paper trading with achievements)
openquant game start -s earnings-surge -b 100000

---

Game Mode

Paper trading, but fun. Track your P&L, earn achievements, and compete on the leaderboard.

openquant game start -s value-deep -b 100000

• Start with $100K virtual capital
• Execute trades through any strategy
• Track wins, streaks, and risk management
• Upgrade to Signal Mode (real signals, manual approval) or Live Mode (automated with risk limits)

---

For AI Agents

OpenQuant is designed to be the hands for any AI agent — Claude Code, Hermes, Codex, Cursor.

MCP Server

# Start the MCP server (SSE transport)
openquant-mcp

# 10 tools available:
# openquant_analyze, openquant_strategy_run, openquant_risk_assessment,
# openquant_insider_scan, openquant_backtest, openquant_trade_execute,
# openquant_portfolio_status, openquant_game_status, openquant_trade_history,
# openquant_strategy_list

Connect from Claude Desktop, Cursor, or any MCP client:

{
  "mcpServers": {
    "openquant": {
      "url": "http://localhost:8001/sse"
    }
  }
}

JSON Output

Every CLI command supports --json for programmatic consumption:

openquant --json analyze AAPL | python3 -m json.tool

{
  "command": "analyze",
  "ticker": "AAPL",
  "data": {
    "company_info": {"name": "Apple Inc.", "sector": "Technology"},
    "prices": [{"date": "2026-04-21", "close": 198.42, "volume": 52841300}],
    "signals": {
      "InsiderSentiment": {"score": 0.72, "direction": "bullish"},
      "Momentum": {"score": 0.18, "direction": "neutral"}
    },
    "risk": {"volatility": 22.4, "sharpe_ratio": 0.99, "var_95": -2.0}
  },
  "timestamp": "2026-04-22T14:30:00Z"
}

Skill Files

Drop-in files for popular AI tools:

• CLAUDE.md — Claude Code instructions (MCP tools, trading workflow, risk rules)
• AGENTS.md — Codex/OpenCode instructions
• .cursorrules — Cursor AI rules
• SKILLS/hermes.md — Hermes agent skill (signal interpretation, position sizing)

---

QuantFetch Integration

OpenQuant works out of the box with free data (yfinance + SEC EDGAR). For production-grade data, connect QuantFetch:

	Free (yfinance)	QuantFetch Pro
Stock prices	~800 tickers, 5yr history	8,000+ tickers, 30yr history
Financials	Basic income/balance	Full XBRL with line items
Insider trades	Limited	All Form 4 filings, real-time
SEC filings	Basic metadata	Full-text search, section parsing
Earnings	Basic EPS	Surprises, guidance, consensus
Crypto	—	BTC, ETH, SOL + 50 more
Speed	Rate-limited	100 req/day free, unlimited Pro
Cost	$0	$29.99/mo

# Set your QuantFetch API key
export QUANTFETCH_API_KEY=qf_demo_key_2026

# All commands automatically use QuantFetch when key is set
openquant analyze AAPL

Get a free API key at quantfetch.ai — 100 requests/day, no credit card.

---

Risk Engine

Built-in risk management that doesn't let you blow up:

• Value at Risk (VaR) — 95% and 99% confidence intervals
• Monte Carlo simulation — 10,000 path projections
• Kelly criterion — Optimal position sizing (with 0.25x conservative multiplier)
• Max drawdown — Historical worst-case tracking
• Beta — Market correlation for hedging
• Stop-loss guards — Automatic position limits

openquant risk AAPL
# Volatility: 22.4%  |  Max Drawdown: 13.8%  |  Sharpe: 0.99
# VaR 95%: -2.0%  |  Beta: 1.24  |  Rating: MODERATE

---

Insider Monitor

Real-time scoring of insider Form 4 filings:

openquant insider AAPL

• Scores each filing: buy vs sell, size, officer vs director
• Aggregates 30-day insider sentiment score
• Flags cluster buying (multiple insiders buying simultaneously)
• Cross-references with price action for confirmation

---

Architecture

src/openquant/
├── agent/          # Dual-loop agent (20-turn limit, streaming, risk hooks)
├── agents/         # 5 analysis agents (insider, momentum, value, earnings, analyst)
├── brokers/        # Paper, Alpaca, Kalshi execution
├── cli/            # Click-based CLI (--json, Rich output)
├── data/           # Pluggable data protocol (yfinance, SEC, QuantFetch)
├── game/           # Game mode engine (achievements, leaderboard)
├── insider/        # Insider scoring (Form 4, cluster detection)
├── mcp/            # MCP server (SSE transport, 10 tools)
├── risk/           # Risk engine (VaR, Monte Carlo, Kelly, sizing)
├── strategies/     # 4 built-in strategies + framework
└── tui/            # Textual TUI (watchlist, chat, portfolio panels)

---

Install

pip install openquant-cli

That's it. No API keys required. Works immediately with free data sources.

With QuantFetch (optional)

export QUANTFETCH_API_KEY=your_key_here

Get a free key at quantfetch.ai.

---

Comparison

Feature	OpenQuant	virattt/ai-hedge-fund	OpenBB
Free data	yfinance + SEC EDGAR	Requires API key	Requires API key
Risk engine	VaR, Monte Carlo, Kelly	None	Basic
Strategy framework	4 built-in + custom	Single analysis script	None
Paper trading	Game mode + achievements	None	None
Live execution	Alpaca, Kalshi	None	None
MCP server	10 tools	None	None
AI agent support	JSON, MCP, skill files	CLI only	SDK
Insider monitoring	Real-time scoring	Basic agent	Basic
--json output	All commands	None	Partial
License	MIT	MIT	MIT

---

Disclaimer

This is an educational and research tool. It is not financial advice. Trading involves substantial risk of loss. Past performance does not guarantee future results. Always do your own research and never trade money you can't afford to lose.

---

Built by Mitchell Bernstein · Powered by QuantFetch

Get a free API key · Report a bug · Contribute