The study of accelerated gradient methods in Riemannian optimization has recently witnessed notable progress. However, in contrast with the Euclidean setting, a systematic understanding of acceleration is still lacking in the Riemannian setting. We revisit the \emph{Accelerated Hybrid Proximal Extragradient} (A-HPE) method of \citet{monteiro2013accelerated}, a powerful framework for obtaining accelerated Euclidean methods. Subsequently, we propose a Riemannian version of A-HPE. The basis of our analysis of Riemannian A-HPE is a set of insights into Euclidean A-HPE, which we combine with a careful control of distortion caused by Riemannian geometry. We describe a number of Riemannian accelerated gradient methods as concrete instances of our framework.