Interactive Tear Sheets for Cointegration and Mean-Reversion Analysis
Summary
This module describes an interactive dashboard for examining two-asset pairs through cointegration tests and mean-reversion models. Its analytics include Engle–Granger and Johansen portfolio construction, augmented Dickey–Fuller results, cointegration vectors, portfolio returns, and spread residual diagnostics. The diagnostics cover dispersion, estimated mean-reversion half-life, skewness, kurtosis, normality, quantile plots, and autocorrelation patterns.
The dashboard also fits an Ornstein–Uhlenbeck model to spread data and presents optimal entry and liquidation levels under user-specified discount rate, transaction cost, and optional stop-loss assumptions. These features offer a workflow for inspecting candidate pairs and model-derived trading levels. The document is implementation code rather than a trading study: it reports no empirical results, comparative validation, or live-trading evidence. Statistical tests and fitted levels depend on the supplied data and modeling assumptions, so the visual outputs alone do not establish that a pair will remain cointegrated or that a strategy will be profitable.
Key ideas
- The dashboard compares two assets using Engle–Granger and Johansen cointegration methods.
- It builds portfolio returns from estimated cointegration vectors.
- Residual diagnostics include half-life, distribution shape, normality, and autocorrelation measures.
- An Ornstein–Uhlenbeck model estimates entry and liquidation levels under configurable costs and stop-loss assumptions.
- The code presents analytics but supplies no evidence of trading performance.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.