The variable forgetting factor-based local average model algorithm for prediction of financial time series

In this paper, we propose a variable forgetting factor-based local average model for estimation of future values of financial time series. The forgetting factor is applied to the existing local average model to govern the weights of past records for the estimation of the future records. By using the trend direction from the turning points of the financial time series, the value of the forgetting factor can be estimated. The results of performance comparison between the proposed variable forgetting factor-based local average model and the original local average model on the actual time series derived from the stocks listed in the Stock Exchange of Thailand are shown. The results suggest that the proposed method offers consistent less prediction errors than the existing method.