The article explores how different characteristics of stocks can help predict their returns in the Chinese market. By using nonlinear predictive models, the researchers found that certain characteristics can provide valuable information for predicting stock returns. They discovered that a portfolio based on these predictions had a higher Sharpe ratio compared to a linear model. The study also looked at how factors like firm size, earnings-to-price ratio, and turnover interact with predictive information. Overall, the findings suggest that nonlinear models can be more effective in predicting stock returns in China.