Statistical and economic evaluation of forecasts in electricity markets: beyond RMSE and MAE

Battle Count: 63

Highly relevant to quantitative trading in electricity markets. The paper directly addresses how forecast quality metrics relate to trading profits in BESS arbitrage strategies. It challenges the conventional use of RMSE/MAE for forecast selection in trading contexts and proposes alternative metrics (Corr-f, MPD, Cov-e) that better capture economic value. Applicable to energy trading desks, storage operators, and quantitative analysts working in electricity markets. Less directly applicable to traditional equity/bond quantitative trading.

Moderate complexity. The forecasting models (ARX, NARX, LEAR) are standard and well-documented in literature. The BESS arbitrage strategy is simplified (single cycle, block-based charging/discharging). The main complexity lies in constructing the 192-forecast pool with multiple specifications, transformations, and window sizes. The evaluation metrics (Cov-e, Corr-f, MHD, MPD) are straightforward to compute. No specialized optimization solvers required due to simplified trading rules.

Data sources are publicly available (ENTSO-E transparency platform, Investing.com). Model specifications are well-documented with clear parameter choices. The forecasting pool construction is described in detail. However, no code repository is explicitly mentioned. The BESS trading strategy is simplified but clearly defined. The 5-year out-of-sample period (2020-2024) is specified.