ANALYSIS OF COVID-19 IMPACT ON INDONESIAN SECTORAL STOCKS WITH WAVELET NEURAL NETWORK (WNN) METHOD

• Main Author: Adila Balqis, Sarah
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/52951
• Institution: Institut Teknologi Bandung

Econophysics is a branch of complex systems where problems in the economic field are solved by the principles and methods that exist in physics. The application of econophysics could be used to analyze the prediction of changes in stocks prices in the capital market. Stocks are one of the economic instruments that most commonly traded for investment purposes in the capital market.The methods that can be used to predict stock prices is the Wavelet Neural Network (WNN). This Method is a combination of the wavelet and the artificial neural network methods. In this method, wavelets are used to denoise stocks data before being modeled using neural network. This study uses Daubechies wavelets of order 2 with decomposition level 2 and 3, then used Long-Short Term Memory (LSTM) which is a type of Recurrent Neural Network (RNN). This study aims to determine the effect of Covid-19 on sectoral stocks in Indonesia. The data used is sectoral stock data from March 2018 - July 2020 which is divided into train data, test data, and prediction data. The train and test data used are the data before the Covid-19 case appeared in Indonesia, that is March 2018 - February 2020. The predictive data used is from March 2020 - July 2020 which is period after the Covid-19 case appeared in Indonesia. The prediction results using the data denoised by the Daubechies of order 2 decomposition level 2 wavelet, gave the best accuracy in the trade, services and investment sectors at 98,26% and the smallest accuracy value was 96,45% in the financial sector. While the prediction results using the data denoised by Daubechies order 2 decomposition level 3 wavelet gives the largest accuracy value of 98,13% in the trade, services and investment sectors and the smallest accuracy value was 96,15% in the financial sector.