The article presents a new test to compare nonparametric regression curves without needing a specific smoothing number. The test is based on a simple calculation and works well even when the data distributions are different. It can detect differences between curves that converge to a common point slowly. The test has been shown to be effective in small samples, with good accuracy and power. The researchers also explored how to apply the test to multiple regression and comparing several regression curves.