We assess the Flare Anticipation Index (FAI), a GOES-based thermal precursor to flare onset, on a representative multi-cycle sample. Using 200 evenly spaced days between 1980–2025, we compute temperature and emission measure from GOES/XRS and generate alerts based on a specific FAI parameter setting, then associate alerts with GOES events within a 30-min window following. Across 1,137 flares, 84.5% show
$\geq$1 prior FAI alert; the detection fraction rises with class (C/M/X) and is weaker for A/B events. Viewed from the alert side, of 16,161 total FAI alerts, 42.8% are associated with a flare, while 57.2% remain unassociated (candidates for false positives). A case study (M2.1, 2024-02-05) shows
$\sim29$ min lead time. Flux-EM correlations at peak support the thermal-buildup interpretation. These preliminary results indicate that the FAI is sensitive to medium/large flares and potentially useful for real-time nowcasting, while highlighting limitations for weak events and the need to reduce false positives through thresholding and window optimization.