All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Advanced Quantitative Indicators: Hurst Exponent, Fractional Differencing, Ornstein-Uhlenbeck (OU) process estimation.
- Kalman Filter implementation for signal smoothing and state estimation.
- Black-Litterman Portfolio Optimization model.
- Unified single-package architecture for streamlined usage and improved performance.
- Automated CI/CD workflow with Node 22 support.
- Major Structure Reductions: Consolidated 12 independent packages into a single unified meridianalgo package.
- Migrated all source code from individual packages/*/src to root src/.
- Updated all internal imports to relative paths.
- Rebranded as meridianalgo (removing @meridianalgo/ scope for simplicity in the unified version).
- Improved Technical Indicators with consistent NaN padding and robust empty-array handling.
- Fixed Bollinger Bands calculation errors where bands were misaligned with the middle line.
- Corrected HMM regime detection crash on short datasets.
- Adjusted risk metrics conventions to match industry standards (negative drawdowns).
- Complete monorepo architecture with 15 packages
- Plugin system for extensibility
- TypeScript support with full type definitions
- Comprehensive test coverage
- Yahoo Finance adapter (fully functional)
- Polygon.io adapter with rate limiting
- Binance adapter with WebSocket streaming
- Alpaca adapter with paper/live trading support
- Data manager with caching and quality validation
- Corporate action handling
- Gap filling and normalization
- Classic indicators: SMA, EMA, RSI, MACD, Bollinger Bands, Stochastic, ATR
- Advanced volatility: GARCH, EWMA, Realized Volatility
- Regime detection: HMM, Change Point Detection
- Microstructure: VPIN, Order Imbalance, Kyle's Lambda
- Feature engineering: PCA, Lags, Rolling Statistics
- Seasonality detection: Day-of-week, Month-end, Holiday effects
- Trend Following strategy
- Mean Reversion strategy
- Pairs Trading strategy
- Momentum strategy
- Strategy composition (blend, vote, regime-gating)
- Position sizing (Kelly, Vol Target, Drawdown Aware)
- Trade rules and constraints
- Time-based backtest engine
- Event-driven backtest engine
- Realistic cost models (commission, slippage, borrow fees)
- Walk-forward analysis
- Multi-scenario testing
- Corporate action handling
- Performance metrics calculation
- VaR (Historical, Parametric, Monte Carlo)
- CVaR, Max Drawdown, Volatility
- Sharpe, Sortino, Calmar, Information Ratios
- Performance Attribution (Brinson, Factor Exposure)
- Stress Testing (scenario, historical, Monte Carlo)
- Real-time monitoring and alerts
- Reporting system (PDF/HTML tear sheets)
- Mean-Variance Optimization
- Black-Litterman
- Risk Parity
- Hierarchical Risk Parity (HRP)
- Constraint handling (min/max weights, long-only, leverage, sector limits)
- Robustness techniques (covariance shrinkage, resampled frontier)
- Multi-objective optimization
- Random Forest, Gradient Boosting
- Neural Networks, LSTM, GRU
- ARIMA, VAR, Kalman Filter
- Hyperparameter tuning (grid search, random search, Bayesian)
- Online learning with drift detection
- Feature selection and importance
- AutoML system
- Labeling methods (triple-barrier, meta-labeling)
- Paper trading (fully functional)
- Live trading (Alpaca integration)
- Order Management System
- Pre-trade risk checks
- Risk limits enforcement
- Smart order routing
- Bracket orders
- Compliance logging
- Grid search
- Random search
- Bayesian optimization
- Genetic algorithms
- Parallel evaluation
- Overfitting prevention
- Result visualization
- Equity curves
- Drawdown charts
- Return distributions
- Correlation heatmaps
- Factor exposure charts
- Dashboard system
- Export functionality (CSV, JSON, PNG, SVG, PDF)
- DAG execution
- Feature store with versioning
- Scheduling support (cron, event-triggered)
- State management (SQLite, PostgreSQL, file-based)
- Pipeline monitoring
- Audit logging
- Run management
- Artifact tracking
- Policy enforcement
- Snapshot system
- Regression testing
- CLI for common tasks (init, backtest, optimize, live, report)
- Project templates
- Configuration validation (Zod)
- Progress reporting
- Comprehensive documentation
- Code examples
- Math utilities (mean, std, correlation, percentile, skewness, kurtosis)
- Statistical tests (t-test, chi-square, ADF, Jarque-Bera, bootstrap)
- Time utilities (market hours, trading days, holidays, resampling)
- Structured logging with trace IDs
- Migrated from single package to monorepo architecture
- Improved TypeScript type safety
- Enhanced error handling
- Optimized performance for large datasets
- Various bug fixes and improvements
- Initial release with basic indicators
- Simple backtesting functionality
- Basic performance metrics
For more details, see the documentation.