Understanding the formation of galaxy clusters is important for studying the role environment plays in the evolution of galaxies. At high redshift, z > 1.5, we witness the early stages of cluster formation, and these structures are referred to as protoclusters. The terms cluster and protocluster, while often used interchangeably, do not denote the same thing. Clusters are typically single virialised haloes, while protoclusters are the diffuse collection of haloes that will merge to form the final low redshift cluster. Using a semi-analytic model applied to the Millennium Simulation, I will demonstrate that protoclusters are very extended objects that are not necessarily dominated by a single dark matter halo. Additionally, I will explain the difficulties in predicting the z=0 mass that these structures will evolve into; however, I will also demonstrate that the precision of this mass estimate can be improved when halo pairs are taken into account. Finally, I will show a practical example of where we use our knowledge from simulations to study an observed high redshift cluster.