Summary: In partizan games, where players Left and Right may have different options, there is a partial order defined as preference by Left: \(G \geqslant H\) if Left wins \(G+X\) whenever she wins \(H+X\), for any game position \(X\). In normal play, there is an easy test for comparison: \(G \geqslant H\) if and only if Left wins \(G-H\) playing second. In misère play, where the last player to move loses, the same test does not apply – for one thing, there are no additive inverses – and very few games are comparable. If we restrict the arbitrary game \(X\) to a subset of games \(\mathcal{U}\), we may have \(G \geqslant H\) “modulo \(\mathcal{U}\)”; but without the easy test from normal play, we must give a general argument about the outcomes of \(G+X\) and \(H+X\) for all \(X \in \mathcal{U}\). In this paper, we use the novel theory of absolute combinatorial games to develop recursive comparison tests for the well-studied universes of dicots and dead-ending games. This is the first constructive test for comparison of dead-ending games under misère play, using a new family of end-games called perfect murders.