Electronic Theses and Dissertation

Peramalan harga saham merupakan salah satu upaya untuk membantu pengambilan keputusan investasi berdasarkan pola data historis. penelitian ini bertujuan membandingkan kinerja metode autoregressive integrated moving average (arima) dan long short-term memory (lstm) dalam meramalkan harga saham pt bank central asia tbk (bbca) serta menentukan metode yang memiliki tingkat akurasi terbaik. penelitian ini menggunakan pendekatan kuantitatif dengan data sekunder berupa harga penutupan saham bbca periode januari 2020–maret 2026 yang diperoleh dari yahoo finance. pemodelan arima dilakukan melalui uji stasioneritas, identifikasi model menggunakan acf dan pacf, pemilihan model berdasarkan nilai aic, serta uji diagnostik. sementara itu, pemodelan lstm dilakukan melalui preprocessing data, normalisasi, pembagian data training dan testing, serta optimasi parameter. kinerja kedua metode dievaluasi menggunakan mean absolute percentage error (mape). hasil penelitian menunjukkan bahwa kedua metode mampu melakukan peramalan harga saham bbca, namun metode arima memberikan tingkat akurasi yang lebih baik berdasarkan nilai mape yang lebih rendah dibandingkan lstm. dengan demikian, metode arima lebih sesuai digunakan untuk peramalan harga saham bbca pada data penelitian ini. penelitian selanjutnya disarankan membandingkan kedua metode dengan metode peramalan lainnya atau menggunakan data yang lebih panjang untuk memperoleh hasil yang lebih komprehensif. kata kunci: arima, lstm, harga saham bbca, peramalan, mape.

Stock price forecasting is an important approach to supporting investment decision making based on historical data patterns. This study aims to compare the performance of the Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) methods in forecasting the stock price of PT Bank Central Asia Tbk (BBCA) and to determine the method with the highest forecasting accuracy. This research employed a quantitative approach using secondary data consisting of BBCA's closing stock prices from January 2020 to March 2026, obtained from Yahoo Finance. The ARIMA model was developed through stationarity testing, model identification using the Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF), model selection based on the Akaike Information Criterion (AIC), and diagnostic testing. Meanwhile, the LSTM model was developed through data preprocessing, normalization, training and testing data splitting, and parameter optimization. The performance of both models was evaluated using the Mean Absolute Percentage Error (MAPE). The results indicate that both methods are capable of forecasting BBCA stock prices; however, the ARIMA method achieved better forecasting accuracy by producing a lower MAPE value than the LSTM model. Therefore, ARIMA is considered more suitable for forecasting BBCA stock prices in this study. Future research is recommended to compare these methods with other forecasting techniques or use a longer observation period to obtain more comprehensive results. Keywords: ARIMA, LSTM, BBCA stock price, forecasting, MAPE.