This presentation introduces a multi-resolution spatial grid approach for modeling and evaluating earthquake forecasts using quadtree data structures. Traditional earthquake forecasting experiments commonly rely on fixed-resolution spatial grids, which can produce millions of grid cells, many of which contain no seismic observations. This leads to inefficient model representation, increased computational cost, and reduced statistical robustness in forecast evaluation. The proposed approach uses quadtree-based hierarchical tiling to generate adaptive spatial grids where cell resolution is determined by the spatial distribution of seismicity. Regions with higher earthquake density are represented with finer spatial resolution, while low-seismicity regions remain coarser. The approach enables more efficient forecast generation, improved spatial representation of seismicity, and compatibility with existing forecast evaluation frameworks, including integration with the pyCSEP earthquake forecast testing toolkit.