In this paper, we propose a new polynomial fitting algorithm to improve the spot position detection accuracy based on four-Quadrant Detector (4QD) when the circular spot with Gaussian energy is used as the incident light model. The traditional polynomial fitting method is difficult to ensure high spot position detection accuracy in a wide detection range. To solve this problem, we analyze and compare the characteristics of different algorithms for spot position detection, and consider the influence of the 4QD gap size in the model. Based on the initial solution of the geometric approximation method, we introduce the error compensation factor function, a new spot position detection model is designed. The results of simulation and experiment show that the new algorithm can greatly reduce the position detection error of 4QD for Gaussian spot. When the radius of incident spot is 0.5 mm and within the detection range of [-0.5 mm similar to 0.5 mm], the maximum error is 0.001353 mm and the root-mean-square error is 0.0004596 mm with the new five-order polynomial fitting algorithm which are reduced 56.7% and 69.7% than traditional nine-order polynomial fitting algorithm. Moreover, the computational complexity of the new algorithm is much less than traditional algorithm and the new algorithm also has good prospects in laser communication, high energy laser weapons or others.