According to the present invention, a method by which a decoding device performs video decoding comprises the steps of: parsing a merge index of a current block; configuring a merge candidate list on the basis of neighboring blocks; selecting a merge candidate indicated by the merge index among merge candidates included in the merge candidate list; deriving L0 motion information and L1 motion information of the current block on the basis of the selected merge candidate; deriving an L0 weight and an L1 weight of the current block; and generating a prediction sample of the current block by weighted-summing an L0 reference sample and an L1 reference sample on the basis of the L0 weight and the L1 weight.