Novel time series methods in economic forecasting: SutteARIMA evidence from Indonesian Consumer Price Index and currency exchange rates

Economic forecasting demands precision. Yet traditional models often stumble when confronted with real-world complexity – the messy interplay of linear trends, non-linear disruptions and seasonal fluctuations that characterize financial time series. This study introduces SutteARIMA, a hybrid forecasting approach that marries the α-Sutte Indicator with ARIMA methodology. What makes this interesting? The model does not require users to distinguish between linear and non-linear data patterns beforehand – a practical advantage in volatile economic environments. Testing across 200 pseudo samples revealed SutteARIMA’s versatility across diverse pattern types. The real test came with Indonesian economic indicators during the COVID-19 period. Results speak clearly. For Indonesia’s exchange rate forecasting, SutteARIMA achieved MSE of 66,474.88 and MAPE of 1.33% – outperforming ARIMA by 16%. Consumer Price Index (CPI) forecasting proved even more impressive: MSE of 0.0493 and MAPE of 0.1594%, surpassing α-Sutte by 17%. These performance gains are not marginal; they represent substantial improvements in forecasting accuracy during periods of economic uncertainty. The implications extend beyond technical superiority. Policymakers working with limited preprocessing resources can deploy SutteARIMA confidently across various economic indicators. The method’s consistent performance across both trending exchange rates and stable price indices suggests robust applicability in diverse economic contexts. Economic forecasting during volatile periods presents significant challenges for policymakers and businesses who must choose between linear and non-linear methods before understanding their data characteristics. This research introduces SutteARIMA, a hybrid methodology combining α-Sutte Indicator and ARIMA approaches that automatically adapts to diverse patterns without requiring preliminary classification. Extensive testing across 200 systematically generated pseudo samples representing the full spectrum of time series behaviors—including non-trending, trending, seasonal, and complex mixed patterns applicable across economic, financial, environmental, health, and other temporal data—demonstrates the method's broad versatility. Validation with Indonesian economic data during COVID-19 shows consistent 16-17% performance improvements over traditional methods. For central banks, exceptional CPI forecasting accuracy (MAPE < 0.16%) enables precise inflation monitoring crucial for monetary policy decisions. Exchange rate forecasting improvements benefit government budget planning, central bank interventions, and commercial risk management. The method's modest computational requirements make advanced forecasting accessible to organizations with limited technical resources. This research challenges conventional wisdom requiring practitioners to choose between forecasting paradigms, instead offering an integrated framework that leverages complementary strengths automatically. The demonstrated versatility suggests broad applicability across diverse domains including energy prices, commodity markets, financial indicators, environmental monitoring, epidemiological trends, and any temporal data where pattern uncertainty presents challenges.