Validation of Future Crude Palm Oil Futures (FCPO) Price Prediction

Xhin Rong, Yong and Jais, Bin Mohamad and Tamrin, K. F. (2023) Validation of Future Crude Palm Oil Futures (FCPO) Price Prediction. Masters thesis, UNIMAS.

	PDF Thesis MSc_Yong Xhin Rong - 24 pages.pdf Download (414kB)
	PDF (Please get the password by email to repository@unimas.my , or call ext: 3942 / 3973 / 3933) Thesis MSc_Yong Xhin Rong.ftext.pdf Restricted to Registered users only Download (1MB)
	PDF Yong Xhing Rong _dsva.pdf Restricted to Repository staff only Download (236kB)

Abstract

The implementation of Artificial Intelligence (AI) towards the price prediction process is undoubtedly crucial as the importance of accurate price prediction has increased over the last few decades. The lack of accurate price prediction tools in the past has always been a challenge for investors to predict future prices. Thus, investors tend to use technical indicators to improve their trading strategy. However, with the large amount of different technical indicators in the market, it is tedious and time consuming to identify which technical indicators works the best for them. Crude palm oil plays an important role in the upbringing of Malaysia’s economy. It is the largest Gross Domestic Product (GDP) contributor for the agricultural sector in Malaysia and is exported globally. It is important to predict the price of Crude Palm Oil Futures (FCPO) as it could allow us to hedge against any unforeseen risks and price changes. This study aims to predict FCPO price and optimize various technical indicators using AI. Thus, this study predicts the future price of FCPO using daily historical data such as the opening, high, low, and closing price of FCPO from year 2010 to year 2019 using the Extreme Learning Machine (ELM) model in the Matlab R2020a software. Besides that, various number of hidden neurons and activation function have been used to train and test the ELM model to identify which parameter works best with the technical indicators to provide the most accurate results. The main result shows that the radial basis function has the highest training and testing accuracy as compared to other functions such as the sigmoidal function, sine function and radial basis function when tested against different number of hidden neurons. As for the optimisation of technical indicators, the Williams Percent Range indicator has the highest prediction accuracy as compared to other technical indicators such as Relative Strength Index (RSI), Exponential Moving Average (EMA), Moving Average Convergence and Divergence (MACD) and Stochastic Oscillator. The predictability of FCPO prices using ELM shows that it is possible to predict future prices and thus disputing the weak form of Efficient Market Hypothesis (EMH) that stated it is nearly impossible to predict prices based on historical data. This study could potentially contribute to businesses related to palm oil as business owners can predict future prices and hedge against any unforeseen risks. Not only that, investors and technical analyst can also contribute from this study by identifying technical indicator that has the highest prediction accuracy based on their trading strategy. The implementation of ELM towards the price prediction process can also guide policymakers to adjust their strategies and help them to obtain more agricultural related grants. Keywords: Artificial Intelligence, ELM, FCPO, prediction, technical indicator.

Actions (For repository members only: login required)

View Item