UQ-LOB: Uncertainty-Aware Limit Order Book Mid-Price Forecasting

2026-09-25

AI Summary

This study addresses the absence of reliable confidence indicators in limit order book (LOB) forecasting by proposing the first encoder-agnostic attentive neural process framework for uncertainty quantification. Designed as a lightweight, plug-and-play module compatible with arbitrary LOB encoders, it integrates Gaussian regression with categorical distribution modeling and employs a historical window conditioning mechanism to calibrate predictive confidence. This approach facilitates a paradigm shift from point predictions to probabilistic distribution forecasts, thereby enabling selective trading decisions. Evaluated across 5.2 billion events, the proposed method achieves near-nominal coverage rates and yields directional F1 scores up to 0.88 within high-confidence intervals, substantially enhancing both the reliability and practical utility of LOB forecasting.

Abstract

Forecasting short-horizon mid-price movements from limit order book (LOB) data is central to algorithmic trading, yet most deep LOB forecasters are point predictors: they output a direction or a displacement, but never indicate which of their forecasts can be trusted. We introduce UQ-LOB, a lightweight, encoder-agnostic uncertainty quantification module that attaches to any pretrained LOB encoder and, in the spirit of attentive neural processes, conditions each forecast on a context set of recently completed windows whose outcomes are already realised. The UQ-regression variant outputs a calibrated Gaussian over the future tick displacement, while the UQ-classification variant outputs a categorical distribution over down/up/stationary. Both expose a scalar confidence (predicted signal-to-noise ratio or class probability) that supports selective prediction. On 5.2 billion LOB events across seven cryptocurrency assets and horizons of 5, 10 and 15 seconds, UQ-regression attains near-nominal 68% interval coverage, and restricting to the most confident 10% of predictions raises directional macro F1 by 0.11-0.15 for UQ-regression and 0.05-0.11 for UQ-classification, at every horizon. On large, economically meaningful moves, the tightest confidence tier reaches a directional F1 of 0.88 (down) and 0.83 (up) at the 5-second horizon.

Problem

• Limit Order Book
• Mid-Price Forecasting
• Uncertainty Quantification
• Algorithmic Trading
• Selective Prediction

Innovation

• Uncertainty Quantification
• Limit Order Book
• Selective Prediction
• Attentive Neural Processes
• Mid-Price Forecasting