Skip to content

CHANGELOG.md

Latest commit

 

History

History
217 lines (166 loc) · 8.37 KB

CHANGELOG.md

File metadata and controls

217 lines (166 loc) · 8.37 KB

Changelog

All notable changes to MeridianAlgo will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[6.2.6] - 2026-03-12 "CI/CD Fix Release"

Fixed

  • Fixed CI workflow to run tests on every push
  • Simplified CI pipeline for better reliability
  • Updated PyPI publishing workflow

[6.2.5] - 2026-03-12 "Organization & Automation Release"

Changed

  • Consolidated all documentation into single docs folder
  • Removed duplicate documentation files (DEPLOYMENT.md, RELEASE_NOTES.md, duplicate CHANGELOGs)
  • Removed emoji usage throughout codebase for professional appearance
  • Humanized and professionalized README content
  • Cleaned up unnecessary comments in setup.py
  • Improved code organization and structure

Added

  • New GitHub Action workflow for automatic PyPI publishing on tag push
  • Automated release creation using README as release notes
  • Separate CI/CD pipeline for testing and publishing

Removed

  • Duplicate Sphinx documentation in meridianalgo/docs folder
  • .agent workflow folder
  • Unnecessary summary and duplicate documentation files
  • All emoji characters from documentation

Fixed

  • Linted entire codebase with Ruff formatter
  • Cleaned up code comments and documentation

[6.2.2] - 2026-02-24 "Institutional Stability Release"

Added

  • Production-Grade CI/CD: Comprehensive GitHub Actions workflow with automated ruff linting/fixing, parallel matrix testing, and automated GitHub & PyPI releases.
  • Unified ML Architecture: Reorganized meridianalgo.ml for modular model development. ModelTrainer, LSTMModel, and GRUModel are now first-class exports.
  • Enhanced API Discovery: Major overhaul of __init__.py files across all sub-packages (Core, Risk, Strategies, Execution) for seamless institutional-grade imports.
  • Detailed Technical Documentation: Updated README/Changelog with high-fidelity performance benchmarks and mathematical foundations.

Fixed

  • Critical Import Paths: Resolved systemic ImportError across ML, Strategies, and Core modules.
  • Execution Analytics: Fixed ExecutionAnalyzer integration and NameError in typing.
  • Standardized Exports: Ensured all __all__ lists in sub-modules accurately reflect available functionality.

Changed

  • Version Alignment: Unified package version to 6.2.2 across pyproject.toml, setup.py, and package metadata.
  • API Aliasing: Formalized professional aliases for core quant components (e.g., VaRCalculator, BacktestEngine).

[6.2.1] - 2026-02-24 "Maintenance Release"

  • Internal module reorganization.
  • Bug fixes for portfolio optimization.

[5.0.0] - 2024-11-29 "Advanced Quantitative Development Edition"

Added

Professional Quant Module (meridianalgo.quant)

  • Market Microstructure Analysis

    • OrderFlowImbalance: Order flow imbalance calculations, VIR, weighted pressure
    • VolumeWeightedSpread: Volume-weighted spread, effective spread, realized spread
    • RealizedVolatility: 5-min RV, bipower variation, realized kernel estimator
    • MarketImpactModel: Almgren-Chriss, square-root law, optimal execution schedules
    • TickDataAnalyzer: Lee-Ready algorithm, VPIN, Roll spread estimator, trade duration analysis

  • Statistical Arbitrage

    • PairsTrading: Complete pairs trading strategy with dynamic hedge ratios (OLS, TLS, Kalman)
    • CointegrationAnalyzer: Engle-Granger and Johansen cointegration tests
    • OrnsteinUhlenbeck: OU process modeling, MLE parameter estimation, simulation
    • MeanReversionTester: ADF test, variance ratio test, Hurst exponent
    • SpreadAnalyzer: Spread statistics, Bollinger bands, optimal entry/exit thresholds

  • Execution Algorithms

    • VWAP: Volume-Weighted Average Price execution with participation limits
    • TWAP: Time-Weighted Average Price execution with equal slicing
    • POV: Percentage of Volume execution with dynamic adjustment
    • ImplementationShortfall: Almgren-Chriss optimal execution framework
    • AdaptiveExecution: Adaptive execution based on real-time market conditions

  • High-Frequency Trading

    • MarketMaking: Avellaneda-Stoikov market making with inventory management
    • LatencyArbitrage: Cross-venue arbitrage detection with latency filtering
    • LiquidityProvision: Passive liquidity provision with fill rate optimization
    • HFTSignalGenerator: Order flow toxicity, volume clock returns, noise ratio
    • MicropriceEstimator: Microprice calculation from order book depth

  • Factor Models

    • FamaFrenchModel: Three-factor and five-factor models with t-statistics
    • APTModel: Arbitrage Pricing Theory with PCA factor extraction
    • CustomFactorModel: User-defined factor models with regularization
    • FactorRiskDecomposition: Portfolio risk attribution to factors
    • AlphaCapture: Pure alpha calculation, IC analysis, factor tilting

  • Regime Detection

    • HiddenMarkovModel: HMM for regime detection with Baum-Welch algorithm
    • RegimeSwitchingModel: Markov regime-switching with forecasting
    • StructuralBreakDetection: Chow test, CUSUM test, Bai-Perron test
    • MarketStateClassifier: Composite market state classification
    • VolatilityRegimeDetector: GARCH-based volatility regime identification

Testing

  • Added 200+ comprehensive test cases
  • Integration tests for all quant modules
  • Mock data generators for realistic testing
  • 90%+ code coverage

Documentation

  • Completely rewritten README with professional branding
  • 100+ code examples across all modules
  • Mathematical formulations and references
  • Real-world use cases by professional type

Changed

  • Package Structure: Reorganized for clarity and professional standards
  • Branding: Updated to "Advanced Quantitative Development Platform"
  • Version: Updated to 5.0.0 across all files
  • Test Organization: Consolidated all tests into main tests/ directory
  • Code Quality: Enhanced error handling and parameter validation throughout

Improved

  • Performance optimizations for high-frequency operations
  • Better error messages and user feedback
  • Enhanced type hints and documentation strings
  • More comprehensive logging and debugging support

Removed

  • Unnecessary summary files and documentation duplicates
  • Test files from package directory (moved to main tests/)
  • Temporary demo files
  • Build artifacts and cache directories

Fixed

  • Import path consistency across modules
  • Edge case handling in statistical tests
  • Numerical stability in optimization algorithms
  • Documentation typos and formatting

[4.1.0] - Previous Version

Added

  • Portfolio optimization enhancements
  • Risk analysis improvements
  • ML model updates
  • Enhanced backtesting capabilities

Changed

  • Updated technical indicators
  • Improved API consistency

Performance Benchmarks

All algorithms are optimized for production use:

  • Market microstructure: < 1ms for tick processing
  • Statistical arbitrage: < 100ms for signal generation
  • Execution algorithms: < 10ms per order slice
  • HFT strategies: < 100s for quote calculation
  • Factor models: < 1s for portfolio optimization
  • Regime detection: < 5s for HMM fitting (100 observations)

Migration Guide

From v4.1.0 to v5.0.0

No breaking changes! All existing functionality remains unchanged.

New imports available:

from meridianalgo.quant import (
    PairsTrading, VWAP, MarketMaking,
    FamaFrenchModel, HiddenMarkovModel
)

Or use direct imports:

import meridianalgo as ma
pt = ma.PairsTrading()
vwap = ma.VWAP(...)

Academic References

  1. Almgren, R., & Chriss, N. (2000). "Optimal execution of portfolio transactions"
  2. Avellaneda, M., & Stoikov, S. (2008). "High-frequency trading in a limit order book"
  3. Engle, R. F., & Granger, C. W. (1987). "Co-integration and error correction"
  4. Fama, E. F., & French, K. R. (1993). "Common risk factors in the returns on stocks and bonds"
  5. Hamilton, J. D. (1989). "A new approach to the economic analysis of nonstationary time series"

Support

MeridianAlgo - Advanced Quantitative Development Platform

Built by quantitative professionals, for quantitative professionals.