Foundry: Full Stack

Get the complete Datafye development environment running in under 5 minutes. Includes data, algo runtime, backtesting, and AI assistance.

What you'll have:

• Data Cloud with synthetic market data

• Algo Container runtime with example strategy

• Backtesting engine with sample results

• Environment MCP Server for AI-assisted development

Prerequisites:

• Docker Desktop installed and running

• 12GB+ RAM available

• Python 3.7+ (optional - for testing)

Windows users: You need WSL to install the CLI. Click here for instructions

Setting up WSL for Windows

If you don't already have WSL installed, follow these steps:

1. Install WSL: Open PowerShell or Windows Command Prompt in administrator mode and run:

   wsl --install

   This will install Ubuntu by default.

2. Restart your computer when prompted

3. Launch WSL: Open the Start menu, search for "Ubuntu" (or your chosen Linux distribution), and click to launch

4. Complete initial setup: Create a username and password when prompted

Once WSL is installed and running, continue with the installation instructions below (WSL provides a Linux environment, so follow the Linux instructions).

For more details, see Microsoft's WSL installation guide.

Tip: Consider using Windows Terminal for a better terminal experience.

Step 1: Install the Datafye CLI

curl -fsSL https://downloads.n5corp.com/datafye/cli/latest/install.sh | sudo bash

Verify installation:

datafye

Expected output:

Datafye -- Institutional Grade Trading for Everyone

Datafye CLI is Datafye's command line tool. Try:

datafye help commands        list the top level commands
datafye help <command>       help on a specific command

No sudo access? See CLI Installation for alternative installation methods.

Step 2: Download Deployment Descriptor

curl -o quickstart.yaml https://downloads.n5corp.com/datafye/quickstarts/latest/foundry-full-stack.yaml

This descriptor configures:

• Synthetic dataset (10 symbols with 90 days history)

• Pre-built momentum algo container

• Backtesting engine

• Environment MCP Server

• No API keys required

Step 3: Provision Full Stack Environment

datafye foundry local provision --descriptor quickstart.yaml

Sneak Preview Note: This quickstart shows the expected workflow and interface. During the preview period, provisioning is not yet available and this command will display a "not yet implemented" message. This is expected behavior for the preview.

Expected Output (when available):

🚀 Provisioning Datafye Foundry (Full Stack)...
✓ Validating descriptor
✓ Pulling Docker images
✓ Starting Data Cloud
✓ Starting Algo Runtime
✓ Starting Backtesting Engine
✓ Starting Environment MCP Server
✓ Loading synthetic data

✅ Full Stack ready!

📊 Endpoints:
   Data Cloud:    http://localhost:8080/datafye-api/v1
   Algo Runtime:  http://localhost:8081/algo-api/v1
   Environment MCP: http://localhost:3000

Provisioning takes 5-10 minutes.

What Just Happened?

The deployment descriptor configured:

• Data Cloud - Normalized market data service with REST and WebSocket APIs

• Algo Container Runtime - Managed environment for running trading algorithms

• Backtesting Engine - Historical strategy testing and optimization

• Environment MCP Server - AI-assisted exploration and development

• Synthetic Dataset - 10 symbols with 90 days of historical data

What's running:

• Data Cloud container at http://localhost:8080

• Algo Runtime container at http://localhost:8081

• Environment MCP Server at http://localhost:3000

• Pre-built momentum algo deployed and ready to run

• Backtesting engine ready for strategy testing

About the Synthetic Dataset: The Synthetic dataset attempts to generate realistic market data and is appropriate for demo and API testing purposes. It is NOT suitable for actual algo development - use real market data providers (Polygon, Alpaca, etc.) for developing and validating production algorithms.

Step 4: Verify Services

Option A: Python Test Script

# Download test script (verifies Data Cloud, Algo Runtime, MCP Server, and deployed algo)
curl -o test-full-stack.py https://downloads.n5corp.com/datafye/quickstarts/test-full-stack.py
python3 test-full-stack.py

Output:

✓ Data Cloud is healthy
✓ Algo Runtime is healthy
✓ Environment MCP Server is healthy
✓ Example algo deployed: quickstart-momentum
✓ Historical data: 90 days available

Option B: CLI Commands

# Check all services
curl http://localhost:8080/datafye-api/v1/health
curl http://localhost:8081/algo-api/v1/health
curl http://localhost:3000/health

# View deployed algo
datafye foundry algo list

Step 5: Run a Backtest

Note: Backtesting commands are being finalized. The example shows expected workflow.

# Run backtest (command syntax TBD)
datafye foundry backtest run \
  --algo quickstart-momentum \
  --start 2024-01-01 \
  --end 2024-12-31

Expected output:

🔄 Running backtest...
✓ Backtest complete! ID: bt-20250123-143022

📊 Results:
   Return:     +18.5%
   Sharpe:     1.42
   Max DD:     -12.3%
   Trades:     234

View full scorecard:

datafye foundry backtest show bt-20250123-143022

Step 6: Connect AI Tools (Optional)

The Environment MCP Server at http://localhost:3000 enables AI-assisted development.

Setup:

1. Configure your AI tool (VS Code, Cursor, Claude, etc.) with MCP server endpoint

2. Explore your deployment conversationally

3. Analyze backtest results with AI assistance

See: Integrating with AI Agents

Manage Your Deployment

datafye foundry local stop        # Stop all services
datafye foundry local start       # Restart all services
datafye foundry local deprovision # Remove deployment

Next Steps

Learn the SDK:

• SDK Getting Started - Build custom algos

• SDK API Reference

Deep dive into backtesting:

• Backtesting Your Algo

• Parallelized Backtesting

• Genetic Algorithm Optimization

AI-assisted development:

• Exploring Deployments with AI

• Vibe Coding Algos with AI

Move to trading:

• Trading: Full Stack - Add broker connectivity

Troubleshooting

Docker not running: Start Docker Desktop and retry.

Insufficient resources: Increase Docker memory to 12GB+ in Docker Desktop settings.

Port conflicts: Stop conflicting services or see CLI Reference for custom ports.

MCP server not responding: Check docker logs datafye-mcp-server