In this paper, an Optimized Ambisonic Panning Algorithm (OAPA) which reduces sound localization error, is proposed. In the conventional Ambisonic Panning Algorithm (APA), sound localization is usually different from the panning angle, especially when listeners are not in an ideal listening position, because of low signal separation among other channels. To overcome this problem, an OAPA using window functions is proposed. A proper window function can be verified, comprising of higher harmonic components than 2M+1 and improved DPC and channel separation. Analysis results demonstrate that the proposed method results in higher signal separation among other channels and lower sound localization errors than the conventional APA.