Stock Index Forecasting Model based on Short-term Volatility Trend and KNN Algorithm

Abstract

Stock indices can reflect the overall development of the stock market, which is characterized by many influencing factors and high volatility, and it is a challenging problem to predict stock indices accurately. Accurate prediction of stock indices is important for individual investment benefit, enterprise development and operation, and social and economic stability. In this paper, a stock index prediction algorithm is proposed using the K-nearest neighbor (KNN) algorithm, which combines the short-term index fluctuation trends for several days before the trading day. The experimental results show that the improved KNN algorithm shows better prediction results for several stock indices compared with logistic regression algorithm and traditional KNN algorithm models.