| Abstract [eng] |
The aim of this paper is to research self-similar α-stable symmetric processes. These processes accure in finance market, nature, various physic systems. Processes were modeled by generating data with different stability parameter α, which was changed from 1,1 to 2. Modeled processes were researched with four different methods: R/S statistic, absoliute moment, aggregated variance and variance of residuals method. Hurst index and its standart deviation were estimated for that purpose. Data of different lengths (2000, 5000, 10000) were researched. This research was made repeatedly with data samples of 100 and 1000. To make this research program was written with C++ programming language. Results of Hurst index values were compared with theorical values also mean absolute pecentage error was calculated for each method. Results of this research showed, that results calculated with variance of residuals method and absolute moments method were much closer to theorical values than results with other two methods. |
