Model of Autocorrelative Function of Time Series with Strong Dependence

摘要

The model based on the optimization problem solving to improve the Hurst parameter estimation for time series with long-range dependence is proposed. The model can be adapted depending on the ultimate goal of estimation. The proposed model was tested on artificially generated data with known characteristics and applied to determination of the Hurst parameters of time series of RTS incomes. Development of the new model is actual because of the fact that traditional Hurst parameter estimations [2] may have a long range of values in practical applications due to nonstationary effects.