Long-range correlation properties of stochastic time series y(i) have been investigated by introducing the function σ² _MA = [y(i) - (i)]², where (i) is the moving average of y(i), defined as 1/n y(i - k), n the moving average window and N_max is the dimension of the stochastic series. It is shown that, using an appropriate computational procedure, the function σ _MA varies as n^H where H is the Hurst exponent of the series. A comparison of the power-law exponents obtained using respectively the function σ _MA and the Detrended Fluctuation Analysis has been also carried out. Interesting features denoting the existence of a relationship between the scaling properties of the noisy process and the moving average filtering technique have been evidenced.