Comparison of Stock Prediction Using ARIMA Model with Multiple Interventions of Step and Pulse Functions

Muh. Qodri (Kelompok Keahlian Statistika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Indonesia)
Utriweni Mukhaiyar (Kelompok Keahlian Statistika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Indonesia)
Vira Ananda (Kelompok Keahlian Statistika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Indonesia)
Siti Maisaroh (Kelompok Keahlian Statistika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Indonesia)

Jurnal Ilmiah Sains
Volume 24 Issue 1, April 2024

Stock price predictions based on technical analysis using historical data help investors determine the optimal time to buy or sell shares with the aim of achieving maximum profits. The aim of this research is to compare the results of Kimia Farma's share price predictions using the ARIMA model with intervention analysis of two variables at once, namely the pulse function and the step function. This is the novelty of this research. The data used in this research is daily data on Kimia Farma shares from the period 16 April 2018 to 14 April 2023. The best model produced is ARIMA (0,1,1) with intervention, shown by a MAPE value of 0.3356% and an RMSE of 0.3356%. 4.03. Kimia Farma's share price prediction for the next five days is 906.5548; 905.7875; 905.0206; 904.2542; 903.4882 rupiah. An increase in share prices occurred after the intervention in the period 15 April 2023 to 19 April 2023. Keywords: ARIMA; intervention model; step and pulse function; kimia farma