การทำนายราคาหลักทรัพย์ด้วยวิธีการฟัซซีซัพพอร์ตเวกเตอร์รีเกรสชันและความฉลาดแบบกลุ่ม

Abstract

The stock market is a center for the exchange of securities between companies and investors. The stock market is a way for companies to raise funds from investors or the general public. The company can use the money to expand its business. or invest in various projects. Because of this, the stock market is sensitive. And depends on many factors, whether they are factors outside the country or factors within the country. This research presents a stock price prediction model through factors from fundamental analysis, factors from technical analysis, and factors that influence the stock market, totaling 26 factors, consisting of factors from fundamental analysis, such as the baht exchange rate. Thai to US Dollar, Current Ratio, Quick Ratio, Net Profit Margin Ratio, Total Asset Turnover Ratio, Debt to Equity Ratio Stocks (Debt to Equity Ratio), Earnings per Share Ratio (Earnings per Share Ratio), Return on Equity Ratio (Return on Asset Ratio), Price to Net Profit Ratio (Price Earning Ratio) and Price/Book Value Ratio. Factors from technical analysis include Exponential Moving Average (EMA), Relative Strength Index (RSI), Average Directional Movement Index (ADX), Stochastic. Oscillator %K and Slow Stochastic Oscillator %D and factors influencing the stock market include world gold prices, Brent crude oil prices, WTI crude oil prices, Dow Jones Index, Hang Seng Index, Nikkei Index, highest prices. , lowest price, open price and closing price. Two neural network structures, i.e., 12-2-2 and 18-3-2 are used in this study. The feature extraction is done with the Chaos centroids, and the swarm intelligence algorithm is used in finding the best parameters. For the fuzzy support vector regression (FSVR) in the closing prices prediction. The sample of this study consisted of 20 companies in 6 industry groups. From the Stock Exchange of Thailand, SET50 index group, data is in the period 2018 - 2023. From the experiment, we found that the square root of the mean square error (RMSE) from the blind test is small. The best blind test result is at 0.2431. We can say that the model can predict the stock prices. The factors from fundamental analysis, technical analysis and those that influence the stock market can be used in the stock price prediction.