The Accuracy Measurement of Stock Price Numerical Prediction

• Published in: Journal of physics. Conference series Vol. 1569; no. 3; pp. 32027 - 32032
• Main Authors: Amar, Samsul, Sudiarso, Andi, Herliansyah, Muhammad Kusumawan
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.07.2020
• Subjects: Accuracy; Forecasting; Measurement methods; Numerical prediction; Outliers (statistics); Performance measurement; Physics
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/1569/3/032027

Stock market prediction is both attracting and challenging. The successful of stock price prediction will give financial benefit, therefore, it has attracted many researchers and practitioners since long time ago. One of stock prediction performance measurement is the accuracy. There are many methods for measuring the accuracy, ranging from simple to sophisticated mathematical formulation. A method may be suitable for some type of data of condition to be forecasted while other methods may be suitable for other conditions. This article describes some of those methods. MAPE is the most popular accuracy measure for forecasting since it is intuitive, easy to be interpreted and can be applied to measure the forecasting accuracy for both individual item and across item groups. However, the MAPE has shortcomings. First, the value of MAPE will be undefined if one or more of the actual data are zero. Second, MAPE is very sensitive to outlier data. Third, equal errors above the actual value result in a greater percentage error than those below the actual value. Finally, using the MAPE for forecast methods comparison will lead to a systematically under-forecast result. Even with its shortcomings, MAPE is preferred for the forecasting accuracy method due to its applicability, interpretability and reliability. Thus, it is better to look for ways of correcting the drawbacks of MAPE rather than searching for alternative measures which are less desirable than MAPE. This paper proposes a modified MAPE using moving average of the actual data (MAPEMA) to reduce the drawbacks of MAPE. The result shows that the use of MAPEMA can reduce the problems of using MAPE.