Electric vehicles (EVs) equipped with vehicle-to-grid (V2G) capability can reduce energy costs by charging during low-price periods and discharging during high-price periods. While this decision process can be posed as a static optimisation problem, the presence of uncertain electricity prices and strict user mobility constraints turns it into a challenging dynamic optimisation task. ChargeTrek reframes the single-EV home-charging problem as an Atari-style video game in which the agent perceives time, SoC, and price information through a visual grid representation, later solved using learning-based algorithms. Using real CAISO day-ahead and real-time prices, we benchmark reinforcement learning (C51), imitation learning (DAgger), and rule-based baselines. Our results show that the ChargeTrek achieves up to 36% cost reduction over the stepwise planner and 49% over immediate charging. This gamified formulation offers both interpretability and robustness under uncertainty.