Price forecasting model generated by Deep Learning Artificial Intelligence
A dedicated team of machine learnings specialists has produced our price forecast models - an output of tuning and training algorithms across years of historical market data.
- predispatch and week ahead price forecasts
- refreshed every 1/2 and 1 hour respectively
- forecasts update seconds after new information is published by AEMO
- input data and previous rundatetimes charted
- benchmarking and performance metrics
- download data via API and other sources
Inherent in our price forecasts are likely changes to market behaviour because our price forecast models are based on market fundamentals such as supply, demand and transmission capabilities.
This allows you to anticipate likely market outcomes in advance typical changes to market behaviour such as unit commitments and fuel procurement.
This service is consumed by trading teams, middle office for calculating risk metrics and limits and back office for day and week ahead revenue projections.
Price forecasts - first tier subscription
- Forecast model consumes predispatch and stpasa AEMO data
- Half hour resolution for both predispatch and weak ahead
- Charts include comparisons with AEMO forecasts and earlier rundatetimes
- Charts of main input variables including earlier rundatetimes
- Decades of Machine Learning (also known as AI) experience
- Price forecast model is the output of training machine learning algorithms
- Model retraining schedule included in knowledge base
Performance metrics
- Performance metrics show absolute mean error since model release
- Filter for time of day, rundatetime period and price outcome range
- Compare with AEMO performance
- Sample count of metric included
- Metrics by region