This paper presents a two-layer torque vectoring control strategy for the Formula SAE racing car of Tsinghua University to enhance steering response, lateral stability and track performance. Firstly, the dynamic model of the existing FSAE car is built as parameters of tires, suspensions, motors and aerodynamics are measured and identified. Secondly, this paper develops a two-layer torque vectoring strategy, the upper-layer direct yaw moment (DYC) controller and the lower-layer torque distribution controller are developed in Simulink. The upper-layer sliding mode control DYC controller calculates the target additional yaw moment according to the target yaw rate based on the two-degree-of-freedom (2DOF) reference model, and the sideslip angle is constrained as well. Then the lower-layer torque distribution controller estimates the wheel vertical load through the lateral acceleration of the vehicle, and selects the torque distribution method with the largest lateral force margin according to the tire friction ellipse to improve stability. Finally, based on Formula SAE circuit, skid pad, double lane change and slalom tracks are built in Carsim, and the effectiveness of the direct yaw control, the torque distribution strategy are verified respectively on vehicles with new/worn tires. Through Simulink-Carsim joint simulation, the results prove that the steering response, stability and track performance of the racing car are improved.