GARCH(1,1) suggested for adaptive risk
- MQL5’s CodeBase published “Institutional GARCH (1,1) Volatility Forecaster” on April 17, pitching a MetaTrader 5 indicator as a replacement for Average True Range. - The post says its default parameters use alpha 0.09 and beta 0.90 to forecast next-candle variance and adjust stops or position size. - The idea extends a broader MQL5 push into econometric volatility tools, including earlier GARCH tutorials and libraries. (mql5.com)
Volatility is the market’s speedometer, and most retail trading systems still read it by averaging old price moves with Average True Range. GARCH tries to estimate the next burst instead. (mql5.com) That idea got a fresh retail trading wrapper on April 17, when MQL5’s CodeBase published “Institutional GARCH (1,1) Volatility Forecaster” for MetaTrader 5. The indicator says it can replace Average True Range in stop-loss and position-sizing logic. (mql5.com) The post describes GARCH, short for Generalized Autoregressive Conditional Heteroskedasticity, as a way to model volatility clustering — the pattern where large moves are often followed by large moves. Instead of a fixed-window average, it recursively updates conditional variance from recent shocks and past variance. (mql5.com 1) (mql5.com 2) In plain terms, Average True Range looks in the rear-view mirror. A GARCH(1,1) model tries to infer whether the road ahead is getting rougher before the average catches up. (mql5.com 1) (mql5.com 2) The new MQL5 indicator says it uses default parameters of alpha 0.09 and beta 0.90 and forecasts variance for the next execution candle. Its author says traders can use that output to shrink lot size or widen stops before a volatility spike. (mql5.com) The implementation pitch is also practical: the page says the calculations run natively inside the MetaTrader environment, without external Python dependencies. For retail algo users, that means the model can sit directly inside an Expert Advisor or indicator workflow. (mql5.com) (docs.mql4.com) MQL5 has been building toward this for more than a year. It published an article on January 6, 2025 on forecasting volatility with GARCH, then followed on January 14, 2026 with a broader volatility-modeling library designed to resemble Python’s arch package. (mql5.com 1) (mql5.com 2) The sales pitch on the new indicator is stronger than the evidence shown on the page. In the comments, one user said the tool showed “no forecast whatsoever” and appeared to lag even against Average True Range. (mql5.com) That leaves the real takeaway narrower than the marketing. Retail algo developers are being nudged toward conditional-variance models for risk control, but whether this specific indicator beats simpler heuristics still has to be tested in live or historical trading data. (mql5.com) (mql5.com)
