In this study, the forecasting accuracy of a new forecasting method that is alternative to two major forecasting approaches: exponential smoothing (ES) and ARIMA, will be evaluated. Using the results from the M3-competition, the forecasting performance of this method will be compared to not only these two major approaches but also to other successful methods derived from these two approaches with respect to simplicity and cost in addition to accuracy.

Intermittent demand forecasting is crucial for firms and commercial activities. Recently, many researchers have focused on forecasting methods for intermittent demand and proposed various forecasting techniques. The most prominent methods among these proposed techniques are the Croston method, which is based on exponential smoothing, and its two popular variations: SBA (Syntetos-Boylan Approximation), SBJ (Shale-Boylan-Johnston Approximation). Croston method is widely used in forecasting of intermittent demand and inventory (stock) control. Since these demands usually include zero values, using the ground breaking method developed by Croston in this data becomes inevitable. Nevertheless, there are some shortcomings to this method such as producing biased forecasts and for this reason its variations have been proposed. ATA method is a recently developed forecasting method which is an alternative to exponential smoothing. In this paper we propose a modification of ATA method that can be used for forecasting of intermittent demand. We will compare the results of the proposed approach to those of Croston and other forecasting methods used for intermittent demand forecasting.

This study proposes a hybrid forecasting framework that integrates the Auto-Regressive Integrated Moving Average (ARIMA) model with multiple variations of the Ata model, using a performance-based model selection strategy to enhance forecasting accuracy on the M3 and M4 competition datasets. For each time series, seven versions of the Ata model are generated by adjusting level and trend parameters, and the version with the lowest in-sample symmetric mean absolute percentage error (sMAPE) is selected. To improve robustness and prevent overfitting, the median-performing Ata model is also included. These selected models’ forecasts are then combined with ARIMA outputs through optimized weighting schemes tailored to the characteristics of each series. Given the varying frequencies (e.g., yearly, quarterly, monthly, weekly, daily, and hourly) and diverse lengths of time series, a grid search algorithm is employed to determine the best hybrid combination for each frequency group. The model is applied in a series-specific manner, allowing it to adapt to different seasonal, trend, and irregular patterns. Extensive empirical results demonstrate that the hybrid model outperforms its individual components and traditional benchmarks across all frequency categories. It ranked first in the M3 competition and achieved second place in the M4 competition based on the official error metric, the sMAPE and Overall Weighted Average (OWA), respectively. The results highlight the framework’s adaptability and scalability for complex, heterogeneous time series environments.