OrderFusion+: Probabilistic Buy–Sell Price Trajectory Forecasting in Intraday Electricity Markets

Battle Count: 0

Highly relevant for quantitative trading in energy markets. It provides probabilistic forecasts of price trajectories (not just point estimates) using high-frequency orderbook data, which is critical for executing intraday trading strategies, arbitrage, and risk management in continuous electricity markets.

The model architecture involves custom cross-attention mechanisms, dynamic masking layers, and multi-head quantile prediction. While the code is open-source, implementing the specific dynamic masking logic and handling high-frequency orderbook data requires significant engineering effort.

The paper provides an open-source implementation link (https://runyao-yu.com/OrderFusion/), detailed model configurations, hyperparameter search spaces, and hardware specifications. The data source is identified as commercial EPEX SPOT data, which is expensive but standard for this domain.