Electricity price forecasting · NEM
We see electricity prices before the market does.
VoltSight turns research-grade forecasting into trading and dispatch decisions that pay.
The problem
Wholesale prices don't move. They detonate.
Typical hour
$74/MWh
This week peaked at
$1,177/MWh
Swing
16×
Australia's National Electricity Market reprices every five minutes. One evening of scarcity can out-earn a month of calm — for whoever saw it coming.Settled AEMO spot prices, QLD1 — Winter week, 12 – 18 Aug 2023, 30-minute intervals.
The stakes
Trade that week on the official forecast, and a single battery leaves $1,918 on the table.
Storage, traders and flexible loads all act on a forecast. When the forecast misreads a price spike, the money simply goes to someone else. (1 MW / 2 MWh battery, one volatile week of QLD market data — see the simulation below.)
The forecast
One week. Two forecasts against the real price.Only one read the evenings right.
What actually happened — the settled spot price.
MAE · ours
$35/MWh
MAE · official
$192/MWh
sMAPE · ours
52%
sMAPE · official
62%
Real QLD1 data, Winter week, 12 – 18 Aug 2023: actual = settled AEMO spot price; official = AEMO predispatch at the matching lead time; ours = the research model over the same week.
The payoff
Same battery. Same week. Different forecast.Different revenue.
Our forecast
Official forecast
── cumulative revenue · ▮ discharge · ▮charge — each battery's schedule
On our forecast
$0
On the official forecast
$0
That's +27% revenue from the forecast alone.
Simulation: 1 MW / 2 MWh battery, 90% round-trip efficiency, optimal arbitrage planned on each forecast, settled at actual prices. Real QLD1 week (Winter week, 12 – 18 Aug 2023): settled AEMO prices, AEMO predispatch as the official 24h-ahead forecast, and our model's predictions at the same lead.
Beyond batteries
One forecasting engine. Every downstream decision.
shown above
Battery arbitrage
Charge low, discharge high — timed by forecasts that see spikes coming.
in development
FCAS bidding
Co-optimize energy and frequency-control markets with price-aware bids.
in development
Demand response
Shift industrial load away from the hours that hurt.
in development
Retail hedging
Price retail contracts on sharper expectations of wholesale risk.
Who we are
Energy-market ML researchers, QUT
Deep-learning price forecasting and battery optimization research, built on AEMO market data — peer-reviewed methods, engineered for the real market.
A research project of Queensland University of Technology (QUT)
Your assets. Our foresight.
We partner with storage operators, traders and retailers to pilot forecasting-driven decisions on real portfolios.
Partner with us