Empirical Analysis for Stock Price Prediction Using NARX Model with Exogenous Technical Indicators

Abstract

Stock price prediction is one of the major challenges for investors who participate in the stock markets. Therefore, different methods have been explored by practitioners and academicians to predict stock price movement. Artificial intelligence models are one of the methods that attracted many researchers in the field of financial prediction in the stock market. This study investigates the prediction of the daily stock prices for Commerce International Merchant Bankers (CIMB) using technical indicators in a NARX neural network model. The methodology employs comprehensive parameter trails for different combinations of input variables and different neural network designs. The study seeks to investigate the optimal artificial neural networks (ANN) parameters and settings that enhance the performance of the NARX model. Therefore, extensive parameter trails were studied for various combinations of input variables and NARX neural network configurations. The proposed model is further enhanced by preprocessing and optimising the NARX model’s input and output parameers. The prediction performance is assessed based on the mean squared error (MSE), R-squared, and hit rate. The performance of the proposed model is compared with other models, and it is shown that the utilisation of technical indicators with the NARX neural network improves the accuracy of one-step-ahead prediction for CIMB stock in Malaysia. The performance of the proposed model is further improved by optimising the input data and neural network parameters. The improved prediction of stock prices could help investors increase their returns from investment in stock markets.

Application of Traditional Machine Learning Models for Quantitative Trading of Bitcoin

“…Khaidem et al [13] consider selecting some financial technical indicators as features to predict the direction of stock prices with the Random Forest. Similarly, some researches choose many popular financial market indicators as independent variables, such as Moving Average Convergence Divergence (MACD), Momentum (MOM), Relative Strength Index (RSI), and so on, to predict stock price changes with machine learning models and get great results [14][15]. In order to reduce the noise from time series data, several researchers apply signal decomposition methods with machine learning regression models on price prediction, and results based on these methods seem to be better than the original one [16][17][18].…”

“…The financial technical indicators, including the increment of the price and volume, Moving Average, Moving Average Convergence Divergence, Momentum, Rate of Change, Relative Strength Index, Stochastic Oscillator, and On Balance Volume, are always referred on the trade of stock markets and could provide some information for investors about the price changes on stocks. Some researchers consider adding these indicators as independent variables to predict the stock price with machine learning models and have satisfactory performance [13][14][15]. Different technical indicators could have different effects on the stock price prediction [14] and a greater number of indicators selection could help to get fewer prediction errors [15].…”

“…Some researchers consider adding these indicators as independent variables to predict the stock price with machine learning models and have satisfactory performance [13][14][15]. Different technical indicators could have different effects on the stock price prediction [14] and a greater number of indicators selection could help to get fewer prediction errors [15].…”

Threats From Unintentional Insiders: An Assessment of an Organization’s Readiness Using Machine Learning

“…This is due to the fact that information goes in both directions within a NARX, whereas within a ANN it only travels in one direction. NARX is superior to other types of recurrent neural networks in terms of its speed of convergence as well as its lower minimum threshold for the number of neurons that need to be calibrated [47]. As a consequence of this, it is superior to other recurrent neural networks in terms of its ability to identify long-term dependencies.…”

“…Organizational data on user actions can be analysed as a time series for patterns of behaviour that may indicate the presence of threats. Unfortunately, ARIMA, a popular machine learning method based on Time series, operates by utilising lags of the differenced series and lags of the projected errors, both of which necessitate the data to be stationary, which is achieved through differencing [47]. ARIMA model is based on the principle of using data with precise values and without any measurement mistakes, which has the advantages of flexibility and better seasonal patterns, but it is less accurate than any other models in time series.…”