MULTIFRACTAL BEHAVIOR IN PRECIOUS METALS: WAVELET COHERENCY AND FORECASTING BY VARIMA AND V-FARIMA MODELS

We introduce a new approach to improve the forecasting performance by investigating the multifractal features and the dynamic correlations of return on spot prices of precious metals, namely, gold and platinum. The Hölder exponent of multifractal time series is employed to detect the critical fluctuations during the financial crises through measuring the multifractal behavior. We also consider co-movement of Hölder exponents and forecast the Hölder exponents of multifractal precious metal time series on coherent time periods. The results indicate that forecasting of multiple wavelet coherence of Hölder exponents of multifractal precious metal time series is efficiently improved by using Vector FARIMA and VARIMA models.