Predicting the returns of the US real estate investment trust market: evidence from the group method of data handling neural network

Abstract Purpose The Group Method of Data Handling (GMDH) neural network has demonstrated good performance in data mining, prediction, and optimization. Scholars have used it to forecast stock real estate investment trust (REIT) returns some countries region, but not the United States (US) REIT market. primary goal this study is predict US market using GMDH then compare its accuracy with that derived from traditional prediction method. Design/methodology/approach To return on index, generalized autoregressive conditional heteroscedasticity (GARCH) model. In test, training samples, testing kernel functions model are controlled investigate their impact machine learning approach. Corresponding experiments were performed GARCH model, accuracies these two approaches compared. Findings Compared GARCH, GMDH’s much higher, indicating approach can provide a highly accurate prices. size samples affect results. particular, function significant accuracy. linear covariance simple train yield predictions, whereas quadratic difficult train. Even small outperform Research limitations/implications Although shows predicting return, still black-box algorithm for financial analysts develop customize. come market, which world’s largest most liquid Social implications This research outperforms forecasting returns. Hence, investors use make more predictions target REITs’ thus better decisions. Future researchers may REITs other markets. Originality/value first apply determine factors performance. For example, discusses network. It also includes short-term daily previously considered, making valuable reference industry analysts.