Volta is a machine learning model trained on more than 4,000 market features, generating price forecasts at 5-minute intervals across a 24-hour rolling horizon.

The model's error rate is roughly half that of AEMO's pre-dispatch, because it picks up on dynamics simpler forecasts can't: generator rebidding behaviour, tighter demand predictions, and the price impact of interconnector flows and constraints. In live operation, the revenue gap opens across three patterns: holding charge for price spikes that traditional forecasts miss, timing charging around the optimal negative prices, and forecasting short spikes that only last one or two intervals.

See below, real benchmarks comparing Volta controlled BESS assets against other high-performing assets. On a revenue per MWh basis, BESS assets utilising the Volta forecast currently rank first in each of their respective states.