A method for video encoding includes encoding prediction information of a current block in a current picture into a coded video bitstream, the prediction information being indicative of a symmetric motion vector difference (SMVD) mode for bi-prediction from a pair of forward and backward reference pictures. The method further includes, when searching for the pair of the forward and backward reference pictures, skipping a reference picture associated with a weighted prediction without checking a picture order count (POC) of the reference picture. The method also includes determining a first reference block for the current block in the forward reference picture and a second reference block for the current block in the backward reference picture, and reconstructing the current block according to the first reference block and the second reference block.