Deep Limit Order Book Forecasting

ArXiv ID: 2403.09267 “View on arXiv”

Authors: Unknown

Abstract

We exploit cutting-edge deep learning methodologies to explore the predictability of high-frequency Limit Order Book mid-price changes for a heterogeneous set of stocks traded on the NASDAQ exchange. In so doing, we release `LOBFrame’, an open-source code base to efficiently process large-scale Limit Order Book data and quantitatively assess state-of-the-art deep learning models’ forecasting capabilities. Our results are twofold. We demonstrate that the stocks’ microstructural characteristics influence the efficacy of deep learning methods and that their high forecasting power does not necessarily correspond to actionable trading signals. We argue that traditional machine learning metrics fail to adequately assess the quality of forecasts in the Limit Order Book context. As an alternative, we propose an innovative operational framework that evaluates predictions’ practicality by focusing on the probability of accurately forecasting complete transactions. This work offers academics and practitioners an avenue to make informed and robust decisions on the application of deep learning techniques, their scope and limitations, effectively exploiting emergent statistical properties of the Limit Order Book.

Keywords: Limit Order Book, High-Frequency Trading, Deep Learning, Mid-Price Prediction, Market Microstructure

Complexity vs Empirical Score

• Math Complexity: 6.0/10
• Empirical Rigor: 7.0/10
• Quadrant: Holy Grail
• Why: The paper employs advanced deep learning models (DeepLOB) and discusses microstructural theory, indicating moderate-to-high math complexity, while also releasing an open-source codebase (LOBFrame) for processing large-scale NASDAQ data and proposing a novel operational evaluation framework, demonstrating high empirical rigor.

  flowchart TD
    A["Research Goal: Forecast Mid-Price<br>Changes in High-Frequency LOB"] --> B["Methodology: <br>Deep Learning Models"]
    A --> C["Data Input: <br>NASDAQ Limit Order Book Data"]
    B --> D{"Computational Process"}
    C --> D
    D --> E["Analyze Forecasting Power<br>vs. Microstructural Characteristics"]
    D --> F["Evaluate Model Utility<br>via Transaction Accuracy Framework"]
    E --> G["Key Findings/Outcomes"]
    F --> G
    subgraph G ["Outcomes"]
        H["Deep Learning shows<br>high prediction accuracy"]
        I["Accuracy does not equal<br>actionable trading signals"]
        J["Proposed operational<br>evaluation framework"]
        K["Released Open-Source<br>LOBFrame library"]
    end