@Demo public class Elliott2005StrategyDemo1 extends Object
Using this example. Shows the explicit builder route: load two stock caches, choose a mean-reversion customization, construct Elliott2005Strategy and register report measures. The coefficient vector defines the spread and the calibration window counts observations. Yahoo data and a desktop plotter are required.
Model and publication. The publication describes a latent mean-reverting Gaussian Markov spread observed with noise, allowing state estimation and comparison with the observed spread. Full-text access was unavailable in this audit; this explanation uses the publication abstract and the implementation, not a line-by-line paper verification.
Implementation in this tutorial. The active customization uses fixed coefficients (1,-1), a long-term-mean reference and entry/exit settings 0.4/0.25 on XOM/CVX with 50 calibration observations. The posterior-state and cointegration variants are commented alternatives. Selecting different customization changes what “fair spread” means; the class name does not guarantee that the active rule reproduces the paper’s estimator.
| Constructor and Description |
|---|
Elliott2005StrategyDemo1() |
public static void main(String[] args)
public void run()
Copyright © 2026. All rights reserved.