Method of calculation of the optimal parameters of a predictive model on the basis of a power polynomial canonical expansion of a random sequence of state changes of a technical object or system is offered. The interval between measurements and number of sampling points, the duration of aftereffect and also the order of probabilistic relation are calculated parameters. The flow chart of the algorithm for determining specified parameters is presented in the work as well. The use of optimal characteristics will allow to take into full consideration the properties of an investigated random sequence and consequently to maximize the quality of solving the problem of individual forecasting the reliability of technical objects.
