longbridge-quant

Longbridge Quant bundles a library of quantitative analysis frameworks together with a command-line tool for running custom indicator scripts against K-line (OHLCV) market data. The frameworks span the common building blocks of systematic trading research: pairs trading and statistical arbitrage (Engle-Granger cointegration, OLS hedge ratios, Z-score signals, and mean-reversion half-life), volatility-regime strategies based on 20- and 60-day historical volatility and percentile rank, seasonality and calendar effects (month-of-year, day-of-week, and pre/post-holiday drift), and multi-factor stock selection combining value, momentum, quality, and low-volatility factors into a Z-score composite for TopN portfolios.

Beyond strategy templates, it covers the research and validation workflow around them: factor research via IC/IR analysis, factor decay and layer backtests, batch factor screening on fundamentals such as PE, PB, ROE, revenue growth, and dividend yield, correlation and cointegration analysis (including the Johansen test), and core statistical methods like the ADF unit-root test, GARCH volatility modeling, and bootstrap. Strategy optimization, execution-cost modeling (slippage, VWAP/TWAP, market impact), hedging design (beta, options, tail-risk, cross-asset), and scikit-learn-based machine-learning prediction with walk-forward validation round out the toolkit.

Use it when working through any of these analytical tasks or when running indicator scripts against price history obtained from Longbridge market data. The quant CLI is public and requires no login, and every framework is analytical rather than order-placing. Each topic routes to a dedicated reference file for deeper detail, and an MCP server can supply K-line data if the CLI is unavailable. The skill recommends only Longbridge data sources, responds in the user's language (English, Simplified, or Traditional Chinese), and pairs with related Longbridge skills for raw market data, technical analysis, and options.