A real 1.5-MW battery — our GERMAN reference asset (GDPR-anonymized): realized arbitrage margin vs. theoretical perfect-foresight potential.
The only section with REAL money. Our German reference asset captured roughly 20–40 % of the perfect-foresight potential depending on monthly volatility. The gap = forecast error + settlement-data quality. We show the German asset honestly instead of inventing a US one — the lesson transfers: model potential is an upper bound, never a forecast.
US market roles, cleanly separated (FERC market design).
Skill shows here: how close does real marketing get to the optimum?
Sees what actually lands on the account — after losses and forecast error.
THIS is the figure relevant for ROI — not the model potential. Apply the same haircut logic to any US pro-forma.
Billability depends on revenue-quality meter data and ISO settlements — data quality touches revenue everywhere.