Dynamic Difficulty Adjustment (DDA) systems procedurally tune video games during runtime to deliver a designer-specified, balanced gameplay experience for players. DDA systems have been well-studied and deployed in both academic and industry contexts. However, a relatively unexplored challenge of DDA systems is how to assess the diversity of experiences they can deliver to players and whether or not the range of possible DDA actions will satisfy the original game designer's goals. DDA systems are inherently unpredictable in their output due to being designed to react to game-states that are uncertain and ever-changing. The varying scope and unpredictability of DDA systems means human playtesting can be time-and cost-intensive, and automated playtesting may produce misleading results. In this work, we introduce an approach for using expressive range analysis and unsupervised clustering to explore and evaluate gameplay traces from an in-development DDA system (FighterDDA) for turn-based role-playing game encounters. We find that this is an effective method for assessing designer goals and re-tuning accordingly an in-development system by visualizing and understanding the character of different DDA approaches. While specific to this system, we believe that there is promise in extending this approach to other genres and DDA platforms in the future.