#TITLE_ALTERNATIVE#

• Main Author: MUHAMMAD NUR (NIM : 10213089), DIMAS
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/26662
• Institution: Institut Teknologi Bandung

<p align="justify">Econophysics is a field that applies the concept of physics to solve problems in the economic field. One of the interesting issues to discuss is stochastic dynamics in stock data. Stock data is volatile and unpredictable. In this research, econophysics will be implemented to predict of stock data for agricultural sector in Indonesia Stock Exchange (BEI) using ARCH-GARCH model and Artificial Neural Network – Backpropagation. These two models are so popular and give the best result for prediction. Then, ARCH-GARCH model will be combined with Backpropagation to make hybrid model (GARCH-Neural Network). The concept of this hybrid model is the result of ARCH-GARCH will be used as input and target of Backpropagation model. Company data in agricultural sector divided into three subsectors are used as samples for this research. They are crops subsector, plantation subsector, fisheries subsector with sample period is from January 2015 until August 2017. Stock Data from January 2015 – April 2017 will be trained for predict data from May 2017 – August 2017. Mean Absolute Percentage Error (MAPE) will be used as criteria of model accuracy of predicted results. Smaller MAPE show better predicted results. In this research obtained MAPE calculation results on stock price prediction for three models, 1) GARCH, 2) Backpropagation, 3) GARCH-Neural Network. Crops subsector with MAPE, 1) 1.174%, 2) 1.816%, 3) 0.820%. Then, plantation subsector with MAPE, 1) 0.822%, 2) 1.175%, 3) 0.342%. Fisheries subsector with MAPE, 1) 2.706%, 2) 4.139%, 3) 0.458 %. Those results show that GARCH-Neural Network gives the better result than GARCH model and Backpropagation model.<p align="justify">