Beat the Counter First: A Baseline for Temporal-Graph Anomaly Models

Progress in streaming, edge-level graph anomaly detection (GAD) has been marked by increasingly elaborate architectures, from count-min-sketch chi-square tests to memory-augmented attention networks. Yet the empirical gains attributable to this added complexity have not been systematically evaluated. We propose SimpleCount, a reference with no learning- or gradient-descent-based parameter fitting that selects one scalar feature per dataset from a fixed pool of counts, recencies, first-occurrence indicators, and count-derived transforms. We compare SimpleCount with two temporal-graph detector models and an IsoForest control fitted to the complete feature vector across five public datasets and one synthetic dataset. SimpleCount matches or exceeds SLADE - a state-of-the-art GAD method - on three of six datasets and exceeds IsoForest on all six. We report paired statistical tests and five-seed SLADE evaluations. SLADE requires 23 to 133$\times$ more wall-clock time than SimpleCount. On synthetic datasets, pre-event structural scores recover the planted signal at AUC up to $0.955$, while all evaluated GAD models remain near random. SimpleCount is not a replacement GAD model, rather, its purpose is to highlight that the benefit of complexity is dataset-dependent, and every claimed gain should be reported against a strong one-feature reference together with its compute cost.