نمذجة وتقدير السلاسل الزمنية المالية باستخدام نموذج الذاكرة الطويلة: سوق فلسطين للأوراق المالية نموذجا

Abstract EN

This study aimed to using long memory model to modeling and estimation of financial time series, through the study and analysis of time series data for stock prices in Palestine stock market, using the autoregressive fractionally integrated moving average model -ARFIMA (p, d, q)-, where we have verified by several mathematical and graphical tests; that the series have a long memory property.

then we determined the value of the fractional difference parameter (d) for arfima model; using three estimating methods, where dSperio method outperformed on the other three methods, R / S, Fracdiff and Empirical Hurst exponent, to obtain the ARFIMA model required.

Statistical results also indicated that the most appropriate model to represent the data series Stock prices in Palestine Stock Market is ARFIMA (2,0.3134419,2), with a value of fractional difference (d = 0.3134419), which has succeeded in all necessary Diagnostic statistical tests.