Law enforcement agencies increasingly adopt algorithms for their policing practices, leading to new accountability challenges. However, academic literature tends to regard algorithmic applications as a single entity, resulting in broad and general statements on these risks. By introducing a sociotechnical model of algorithmic policing, this paper identifies eight accountability challenges that originate from technical, organisational and human components. It is illuminated how algorithmic policing accountability is a process that consists of multiple steps that should be considered separately to scrutinise the accountability challenges. The present study showcases how the sociotechnical context determines whether and how accountability issues might arise in algorithmic policing practices. The accountability challenges are presented as propositions that can direct future research in assessing the accountability risks of algorithmic policing applications.