Diffusion transformer improves short term limit order volume prediction accuracy
OrderFlowDiT: Transport-Regularized Diffusion Transformer for Probabilistic Limit-Order Supply Forecasting
Machine LearningArtificial Intelligence
Summary
Predicting the volume of buy and sell orders in stock trading is important but hard because the order book changes quickly and has many details. The authors created a new method called OrderFlowDiT that uses a special transformer model combined with a mathematical technique called optimal transport. This helps the model better predict not just total volume but also where orders are likely placed at different prices. Their testing on real stock data showed that this approach made predictions more accurate and reliable compared to earlier methods.
What this means in practice
- •For quantitative trading teams: Improve short-term forecasts of order flow volumes and placements to support better automated trading decisions.
- •For market data platform providers: Enhance market data analytics with more accurate probabilistic predictions of limit order volumes and price distributions for client services.$Commercial implications: Allows building enhanced forecasting tools for financial firms that pay for actionable order flow insights.
Authors
Jung Hun Phee, Tucker Balch
Abstract
We study the conditional distribution of limit-order volume submitted over the next five minutes, given the current order book and recent order flow. The task requires capturing price placement, total volume, and predictive uncertainty jointly. Coordinate-wise diffusion losses do not explicitly account for price displacement, while normalizing fields to unit mass removes information about total volume. Agreement on selected summary statistics also does not establish fidelity to the joint field distribution. We introduce OrderFlowDiT, a conditional Diffusion Transformer for unnormalized bid and ask volume-at-price fields, trained with an unbalanced optimal-transport (UOT) auxiliary loss. We also develop a full-field evaluation framework that combines a transport cost sensitive to price placement and total volume with proper scoring rules and calibration diagnostics. In the primary comparison on 1950 validation conditions across 30 U.S. equities, adding UOT reduces full-field transport cost by 45.8% and the transport energy score by 26.1% relative to training with mean squared error alone. Both measures improve for every stock. These findings support incorporating price geometry and volume into generative modeling, with field accuracy and predictive uncertainty evaluated together.