Several time series data consist of fluctuating information such as risks and uncertainties, arising from instability of the series data. The most popular model for these data is Generalized Auto Regressive Conditional Heteroskedasticity (GARCH) model. However, the GARCH model does not capture the influence of each variance in the observation because the model uses long-run average variance. The computation of the long-run average variance only considers on the entire series, so it loses information on different effects of the variances in each observation. This study therefore develops a new forecasting model using fuzzy window variance to replace the long-run average variance to incorporate more recent returns, which will yield greater weight of forecast. The concept of fuzzy sliding window was embedded in GARCH model to capture the influence of each variance in the observation. This study is aimed at improving the effectiveness of forecasting time series, which in turn increases forecast accuracy. A monthly airline passenger volume dataset is used for evaluation purposes. The accuracy of the proposed model is compared with Sliding Window GARCH (SWGARCH) and GARCH. From the results, the proposed model produces forecasts that are almost accurate as the actual data and outperforms the benchmark models. The proposed model is significantly fitted and reliable for time series forecasting. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.