Funkcje kary w modelach regresji: Wpływ na zyski i ryzyko w handlu 
energią elektryczną na rynku dnia następnego
[Penalty functions in regression models: Impact on profits and risk 
in day-ahead electricity trading]

Tomasz Serafin (Doctoral School - Management, PWr, Wrocław)


We propose a novel concept in electricity price forecasting and show that 
minimizing the residual sum of squares when estimating model parameters 
does not lead to economically optimal forecasts. We introduce a custom 
penalty function that is directly related to forecasting accuracy and 
trading losses, and show that forecasts obtained by minimizing this 
function allow to obtain higher Sharpe ratios from trading strategies.