Monte Carlo Volatility Forecasting with Multifractal Random Walks

Summary

The article describes a Monte Carlo layer for a multifractal model of asset returns. It takes fitted parameters, including the Hurst exponent, a distribution choice, and sample volatility, then runs many independent simulated paths over a chosen forecast horizon. Each path’s return volatility is measured, and the resulting values are summarized by statistics such as the mean, median, standard deviation, and confidence interval.

The implementation chooses a cascade depth large enough to cover the requested horizon, generates extra points when the horizon does not match a power of the cascade base, and discards unused returns. Failed simulations are skipped, with statistics computed from successful runs. The article describes an end-to-end MQL5 pipeline and refers to tests for the component methods, but the supplied excerpt omits the detailed statistics code and empirical forecast results. It therefore explains a forecasting procedure and implementation structure rather than establishing predictive accuracy or trading profitability.

Key ideas

• Monte Carlo forecasting estimates a distribution of future volatility by aggregating results from many independently simulated paths.
• The cascade depth is rounded up so the generated path covers the requested forecast horizon.
• Volatility is calculated over the portion of each simulated return series that falls within the horizon.
• Failed simulations are omitted, and the forecast is based on the successful runs.
• The approach describes a model implementation but does not establish that its forecasts are profitable or accurate.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.