AI-driven price forecasting uses advanced machine learning and AI models to predict short-, medium- and long-term wholesale electricity prices with higher accuracy and faster processing than traditional statistical approaches. As power systems become more volatile – due to high renewable penetration, extreme-weather events, storage participation and shifting demand patters – price signals fluctuate more rapidly and are harder to anticipate.  

AI systems can process large volumes of historical, meteorological, market and asset-level data to identify complex, non-linear relationships that conventional approaches may miss. These models may include methods such as gradient boosting, random forests and neural networks, which are widely used to capture complex relationships across multiple forecasting horizons. These models typically combine engineered inputs (e.g. detailed weather forecasts, interconnector flows, outages, congestion indicators) with algorithms designed to capture patterns across multiple time horizons. Many AI tools also generate probabilistic forecasts, helping users understand expected price distributions rather than relying on single point predictions. 

By improving the precision of wholesale market forecasts, AI tools support more informed investment planning, risk management and operational decision making for utilities, traders, renewable developers and storage operators. They help market participants anticipate price spikes, congestion events and renewable-generation swings, strengthening financial performance and improving the bankability of clean-energy projects. 

Main outcomes of the solution 

• Business Value: Enhances risk management and revenue optimization through more accurate price forecasts, enabling more confident renewable energy investment decisions, hedging strategies and storage-dispatch planning. 
• Reliability & Flexibility: Supports more efficient system balancing by helping participants anticipate variability, congestion and renewable intermittency, enabling improved bidding strategies and operational planning. 

Key deployment and scaling factors 

1. Develop robust data pipelines integrating market, weather, demand, generation, transmission and congestion datasets across multiple timescales, supported by strong data quality and governance processes. 
2. Test, back test and validate multiple machine learning algorithms to identify the most accurate and stable method for each market and forecasting horizon, using transparent error metrics and performance evaluation across historical periods. 
3. Implement explainability and transparency tools so users can understand key forecast drivers, build trust in model outputs and support regulatory acceptance. 
4. Ensure interoperability with trading, risk management and operational planning platforms, allowing seamless integration into day-ahead, intraday and futures trading processes. 
5. Establish continuous retraining and performance monitoring processes to maintain accuracy as market fundamentals evolve (renewable penetration, policy shifts, new flexibility resources, weather trends). 
6. Assess regulatory or compliance requirements where AI-based forecasts influence market-facing activities, ensuring alignment with accepted standards for risk and data use. 

Case Study 

Hitachi Energy (2023), Switzerland 

Hitachi Energy launched Nostradamus AI, a machine-learning-based forecasting system designed to improve electricity price predictions across global wholesale markets. According to the International Energy Agency (IEA) Energy & AI Observatory and Hitachi Energy, the system evaluates more than ten machine-learning algorithms and selects the best-performing model for each market and forecasting horizon.[1,2]  It can generate forecasts for a range of energy-specific variables, including system load, wind and solar generation, and market prices, and allows users to integrate their own datasets and parameters to produce customized forecasts. 

A 2023 case study using datasets from CAISO and MISO (United States) and AEMO (Australia) indicates that Nostradamus AI achieved forecast accuracies of up to 93% for selected horizons, based on performance metrics presented by Hitachi Energy within the scope of that case study.[3] 

This case illustrates how AI-based forecasting tools are being applied to compare multiple modelling approaches and produce market-specific price predictions, reflecting growing interest in purpose-built forecasting systems for increasingly variable power markets. 

References:

1. Energy and AI Observatory I IEA (June 2025) 
2. Nostradamus®AI Energy Forecasting Software Solution I Hitachi 
3. Hitachi Energy advances predictability of energy investments forecasting with new AI-powered solution I Hitachi (November 2024) 
4. Nostradamus AI: How Hitachi Secures Renewables in the Grid I AI Magazine (November 2024)