This report presents the design, implementation, and validation of an Agent-Based Model (ABM) developed for long-term price scenario generation of Frequency Containment Reserve (FCR) services in the Nordic power system. The ABM simulates the Nordic day-ahead spot market through 11 meta-agents whose bidding behaviour is learned from 2019 historical data via feedforward neural networks. The model is coupled with a Temporal Fusion Transformer (TFT) that translates the simulated spot prices—together with LMA2021 scenario variables—into long-term price trajectories for FCR-N, FCR-D up, and FCR-D down through 2050. Results for the 2045 scenario horizon show that the ABM-supported TFT produces forecasts that closely track LMA2021 benchmarks while exhibiting substantially lower variance and fewer extreme price events, particularly for FCR-D up. The coupled pipeline is evaluated under the Elektrifiering förnybart (EF) and Elektrifiering planerbart (EP) scenarios, with FCR-N showing the tightest alignment between approaches. Key limitations include the use of pre-pandemic training data, hourly temporal resolution, and potential error propagation across the coupled models.