Recursive estimation for regression with infinite variance fractional ARIMA noise

Recently, there has been a growing interest in modeling financial timeseries using fractional ARIMA models with stable innovations; see, for example, [1]. In this paper, the corresponding nonparametric problem for regression with fractional ARIMA noise is studied. A recursive algorithm for estimating time varying parameters is given. It is also shown that a number of existing algorithms are special case of this proposed algorithm.