Discussing the article: "Building Volatility Models in MQL5 (Part IV): Implementing Long Memory Volatility Processes, FIGARCH, and HARCH"

[MetaQuotes]

Check out the new article: Building Volatility Models in MQL5 (Part IV): Implementing Long Memory Volatility Processes, FIGARCH, and HARCH.

The article delivers MQL5 implementations of FIGARCH and HARCH and updates the volatility library for long‑memory processes. It provides code for Hurst and GPH testing, parameter setup (truncation and horizons), and scripts for fitting, forecasting, and simulations. Readers learn how to apply and compare the models on market data to select an appropriate specification.

Traders and algorithmic developers who rely on the GARCH(1,1) family frequently face an operational dilemma: the model either forgets large regime shocks too quickly or, when tuned for persistence, behaves like an IGARCH that never lets shocks decay. In real markets, volatility often decays much more slowly—clusters can persist for weeks or months—and a single exponential kernel is often a poor approximation. The practical question is therefore concrete: how do you prove that your volatility proxy (|r| or r²) exhibits long memory, and if it does, which MQL5 implementation should you use—FIGARCH with hyperbolic decay or HARCH with explicit multi-horizon aggregation—so that it is both statistically appropriate and operationally viable? 

This article answers that question by: providing formal tests (Hurst R/S and GPH) with sample-size safeguards, offering a visual spectral diagnostic, delivering working MQL5 implementations and demo scripts so you can test, select, fit, and deploy long-memory volatility models in your trading stack.

Author: Francis Dube