Perbandingan Akurasi Pemodelan Volume Perdagangan Saham Menggunakan Metode ARIMA dan Markov Switching Autoregressive (MSAR)

Najmi, Rafi Naufal (2025) Perbandingan Akurasi Pemodelan Volume Perdagangan Saham Menggunakan Metode ARIMA dan Markov Switching Autoregressive (MSAR). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211083-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk membandingkan akurasi pemodelan volume perdagangan saham harian menggunakan dua pendekatan deret waktu, yaitu Autoregressive Integrated Moving Average (ARIMA) dan Markov Switching Autoregressive (MSAR). Data yang digunakan merupakan volume perdagangan saham harian di Indonesia dari 4 Januari 2010 hingga 28 Februari 2025, dengan total 3.955 observasi. Setelah dilakukan pra-pemrosesan terhadap 263 nilai hilang dan 84 nilai ekstrem menggunakan metode imputasi rasio, data dibagi menjadi data pelatihan dan data pengujian berdasarkan periode waktu, yaitu 4 Januari 2010 hingga 31 Januari 2025 untuk data pelatihan, dan 3 Februari hingga 28 Februari 2025 untuk data pengujian. Hasil analisis menunjukkan bahwa model ARIMA terbaik, yaitu ARIMA〖(3,1,1)(0,1,2)〗^5, menghasilkan nilai Mean Absolute Percentage Error (MAPE) sebesar 13,04%, dengan karakteristik prediksi yang cenderung stabil dan mengikuti tren umum. Namun, model ini kurang responsif terhadap lonjakan ekstrem dan volatilitas jangka pendek, serta tidak memenuhi asumsi residual berdistribusi normal, meskipun telah memenuhi asumsi residual white noise. Sementara itu, model MSAR terbaik, yaitu MS(2)-AR(1), menghasilkan nilai MAPE sebesar 20,30%. Model ini cukup mampu mengidentifikasi tren jangka panjang, tetapi tidak berhasil menangkap fluktuasi tajam pada data aktual, serta tidak memenuhi asumsi residual white noise maupun distribusi normal. Berdasarkan hasil tersebut, dapat disimpulkan bahwa meskipun model ARIMA memiliki keterbatasan dalam menangkap perubahan ekstrem, model ini tetap lebih akurat dan stabil dibandingkan model MSAR dalam memodelkan volume perdagangan saham harian pada periode studi. Temuan ini memberikan implikasi praktis bagi investor, regulator, aktuaris, dan penyedia layanan transaksi dalam merancang strategi peramalan volume pasar yang lebih efektif.
============================================================================================================================================
This study aims to compare the modeling accuracy of daily stock trading volume using two time series approaches: Autoregressive Integrated Moving Average (ARIMA) and Markov Switching Autoregressive (MSAR). The data used consist of daily stock trading volumes in Indonesia from January 4, 2010 to February 28, 2025, with a total of 3,955 observations. After preprocessing 263 missing values and 84 outliers using the ratio imputation method, the data were divided into training and testing sets based on time periods: January 4, 2010 to January 31, 2025 for training, and February 3 to February 28, 2025 for testing. The analysis results show that the best ARIMA model, ARIMA〖(3,1,1)(0,1,2)〗^5, achieved a Mean Absolute Percentage Error (MAPE) of 13.04%, with predictive results that were relatively stable and followed the general trend. However, the model was less responsive to extreme spikes and short-term volatility, and did not meet the normality assumption for residuals, although it did satisfy the white noise residual assumption. Meanwhile, the best MSAR model, MS(2)-AR(1), produced a MAPE of 20.30%. This model was fairly effective in identifying long-term trends but failed to capture sharp fluctuations in the actual data and did not fulfill the assumptions of residual white noise or normal distribution. Based on these findings, it can be concluded that although the ARIMA model has limitations in detecting extreme changes, it remains more accurate and stable than the MSAR model in modeling daily stock trading volume during the study period. These findings have practical implications for investors, regulators, actuaries, and transaction service providers in designing more effective market volume forecasting strategies.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ARIMA, Modeling, MSAR, Stock Trading Volume,ARIMA, Deret Waktu, MAPE, MSAR, Volume Perdagangan Saham
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis Q Science > QA Mathematics > QA274.7 Markov processes--Mathematical models. Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA76.F56 Data structures (Computer science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Rafi Naufal Najmi
Date Deposited:	04 Aug 2025 08:48
Last Modified:	04 Aug 2025 08:48
URI:	http://repository.its.ac.id/id/eprint/126765

Actions (login required)

View Item