The prediction of price gap anomaly in Chinese stock market: Evidence from the dependent functional logit model

Abstract

How to accurately predict the price gap anomaly is important for asset pricing and risk management. This paper considers the dependence features of high-frequency data and proposes a new dependent functional logit model to predict price gap anomaly, which is an extension of the traditional functional logit model (Escabias et al., 2004). Numerical simulation results demonstrate that our model is significantly superior in terms of out-of-sample prediction accuracy compared with the model that constructed by the FPC estimation and NW estimation. Combining the current day's 5-minute closing price data of the Shanghai Securities Composite Index (SSEC) from Jan 2, 2019 to Dec 31, 2019, we employ this model to predict the next day's price gap anomaly. After comparing with traditional machine learning algorithms, the empirical results find that the prediction effect of our model is better than that of the RBF kernel SVM model and the GBDT model.