ANALISIS PREDIKSI HARGA SAHAM PT. ERAJAYA SWASEMBADA TBK MENGGUNAKAN METODE ARIMA

Abstrak

Perkembangan industri teknologi di Indonesia yang menyebabkan fluktuasi harga saham perusahaan ritel teknologi menjadi semakin dinamis, sehingga diperlukan metode peramalan akurat untuk investasi. Kebaruan penelitian ini terletak pada pemanfaatan Pt Erajaya Swasembada Tbk sebagai subjek kajian ARIMA, mengingat studi pada sector distribusi dan ritel gawai yang terhubung langsung dengan pertumbuhan perangkat digital di Indonesia masih sangat terbatas.  Penelitian ini bertujuan menganalisis karakteristik pergerakan harga saham, menentukan model ARIMA terbaik, serta memprediksi harga saham periode mendatang. pendekatan kuantitatif deret waktu digunakan berdasarkan harga penutupan harian periode Januari 2023–April 2026 (788 observasi) dari Investing.com. Analisis dilakukan melalui uji stasioneritas, identifikasi ACF dan PACF, estimasi model, uji diagnostik residual, dan peramalan via RStudio. Hasil menunjukkan model terbaik adalah ARIMA (1,1,1) dengan nilai MAPE 1,8425% (akurasi sangat baik). Hasil prediksi memperlihatkan harga saham cenderung stabil pada kisaran Rp410 selama 20 periode kedepan. Penelitian ini membuktikan efektivitas ARIMA untuk saham emiten ritel teknologi dan menjadi referensi investor. Penelitian selanjutnya disarankan mengombinasikan ARIMA dengan variabel eksternal atau metode hibrida.

The development of the technology industry in Indonesia has caused stock price fluctuations of technology retail companies to become increasingly dynamic, necessitating accurate forecasting methods for investment purposes. The novelty of this study lies in the use of PT Erajaya Swasembada Tbk as the subject of an ARIMA analysis, given that studies on the distribution and retail sectors for electronic devices—which are directly linked to the growth of digital devices in Indonesia—remain very limited.  This study aims to analyze the characteristics of stock price movements, determine the best ARIMA model, and forecast future stock prices. A quantitative time-series approach was employed using daily closing prices from January 2023 to April 2026 (788 observations) sourced from Investing.com. The analysis was conducted through stationarity tests, ACF and PACF identification, model estimation, residual diagnostic tests, and forecasting via RStudio. The results indicate that the best model is ARIMA (1,1,1) with a MAPE value of 1.8425% (very good accuracy). The forecast results show that stock prices are likely to remain stable around Rp410 over the next 20 periods. This study demonstrates the effectiveness of ARIMA for technology retail stocks and serves as a reference for investors. Further research is recommended to combine ARIMA with external variables or hybrid methods.