Abstract: Hybrid grid is usually composed of spatial cartesian grid and body-fitted grid, which combine the advantages of cartesian grid and body-fitted grid, and has certain advantages in the grid generation and processing of complex geometries shapes. It has become an important grid generation technique in CFD. In the hybrid grid, there is an interface between the spatial cartesian grid and the body-fitted grid, and the flow field information between them needs to be exchanged by interpolation. This process will introduce extra errors and affect the calculation accuracy of the flow field. To solve this problem, Chen et al. proposed flexible-node difference method (FN-FDM) based on DEER algorithm, which can conduct the difference computation of overlapping grid interface flow field and reduce the errors introduced by interpolation computation. The computation stencil of FN-FDM needs to select the grid points from the point cloud around the central point, and different point selection strategies have different effects on the calculation results. Aiming at the point selection requirement of FN-FDM, this paper proposes several different point selection strategies considering angle, orthogonality, distance and other factors, and verifies them by numerical experiments. Numerical experimental results show that the proposed point selection strategy based on the construction idea of FN-FDM, can obtain stable convergence solutions. The distance between the computation point and the central point has the most prominent influence on the computational accuracy, and is the major influence factor of the point selection strategy.