The objective of this study is to introduce a hybrid smoothing method for accurate and reliable analysis of time series indicators of Kazakhstani agriculture for 1995–2018. The method integrates the moving average with convolutional weights derived from the Fibonacci algorithm, specifically designed for boundary points of the series. This combination ensures smoothness without distortion, while retaining the structural properties of the original data. The research employs OECD time series data on Kazakhstani agriculture for 1995–2018. A 5-point moving average was applied for preprocessing. The central parts of the series were smoothed with the Savitzky–Golay (SG) algorithm and the moving average technique, whereas the boundary (initial and final) values were processed using a custom algorithm with Fibonacci-based weight coefficients. The proposed hybrid model effectively reduces irregularities at the edges while maintaining the overall trend shape. The minimal difference between raw and smoothed data demonstrates both the precision and the practical utility of the method. This approach is applicable to building economic models, forecasting, and strategic decision-making. The practical significance lies in its adaptability to fields where edge data are critical, such as agriculture, ecology, and macroeconomics. For Kazakhstan, where policy decisions rely on historical data, its value is particularly high. The scientific novelty lies in combining moving average smoothing with a Fibonacci-based algorithm for boundary adjustment, offering a new direction in time series analysis.