Alpha Machine

The Alpha Machine is a quantitative research environment for building, testing, and deploying custom trading models and factors.

Code Editor

A full VS Code IDE running in your browser, powered by code-server. Write Python scripts, access market data, build and backtest trading strategies — all within an isolated sandbox.

How It Works

  1. Navigate to Alpha Machine → Code Editor
  2. Your workspace is automatically provisioned on first visit
  3. The VS Code editor loads in an embedded iframe
  4. Write Python code, run it in the integrated terminal
  5. Results appear directly in the terminal or as output files

Available Libraries

pandas — data analysis
numpy — numerical computing
scipy — scientific computing
matplotlib — plotting
scikit-learn — machine learning
statsmodels — statistics
cvxpy — optimization
yfinance — market data

Workspace Isolation

Each user gets a separate, isolated workspace:

  • Files are stored in your personal workspace directory
  • You cannot access other users' workspaces
  • Shared example scripts and data are available via symlinks
  • The workspace persists across sessions

Security Restrictions

For security, the following are blocked:

  • os.system(), os.popen(), os.exec*()
  • subprocess.run(), subprocess.Popen()
  • Network access outside of approved libraries
  • File access outside your workspace directory

You can still import and use all data science libraries normally. Only dangerous system-level operations are restricted.

Build Alpha Model

The Build Model module lets you construct alpha models by combining multiple signals. An alpha model produces a score for each stock — higher scores indicate more attractive stocks.

  • Select factors (momentum, value, quality, etc.) and assign weights
  • The combined score ranks stocks from most to least attractive
  • Use scores to inform portfolio construction or screening

Build Screen / Factor

Create custom screening factors by combining fundamental and technical indicators. Screens filter the stock universe down to stocks meeting your criteria.

  • Define conditions (e.g., P/E < 20 AND Momentum > 0.5)
  • Preview which stocks pass your screen
  • Save screens for use in strategy building

Upload Factors

Upload your own custom factor data. This allows you to use proprietary signals (e.g., alternative data, sentiment scores) alongside the built-in factors.

Uploaded factors are available in the factor model, optimizer, and analytics pages.

Related

→ Factor model documentation→ Using models in strategies→ Portfolio optimization