This paper proposes using the genetic algorithms (GAs) for nonlinear time series prediction. A nesting evolution scheme is designed to evolve the forecasting models. In the outer evolution cycle, a binary-coded genetic algorithm is employed to evolve