Risk-aware battery arbitrage optimization using quantile-based price forecast error adjustment

Battery operators participating in electricity markets face uncertain price forecasts that materially affect arbitrage performance, particularly under volatile conditions. This study introduces a risk-aware optimization framework that adjusts pre-dispatch price signals using quantile-based historical forecast-error bounds. By leveraging the 5th and 95th percentiles of past forecast errors, the model introduces dynamic price thresholds to guide charging and discharging decisions under uncertainty. The framework is further extended through a lightweight real-time dispatch layer that preserves a small state-of-charge reserve and enables targeted interventions under cap-proximate or high-volatility conditions, improving responsiveness to unforecast scarcity events. The approach is evaluated on 12 representative trading days from the Australian National Electricity Market in 2025 and further validated over all trading days in June 2025. Performance is benchmarked against both a perfect-foresight strategy and a baseline strategy relying solely on pre-dispatch prices. Across the benchmark set, the baseline pre-dispatch strategy captures approximately 55% of perfect-foresight arbitrage value, while the hybrid framework integrating risk-aware optimization with the real-time dispatch layer increases this capture rate to approximately 60–61% and reduces average CVaR 5 losses by around 3%. The additional month-long validation provides broader evidence of improved aggregate performance across the market conditions observed in June 2025. The proposed framework offers a practical and computationally tractable decision-support tool for battery operators managing short-term price uncertainty in energy-only electricity markets.