Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Recurrent Neural Networks for Dynamic VWAP Execution: Adaptive Trading Strategies with Temporal Kolmogorov-Arnold Networks

Remi Genet

TL;DR

This work tackles VWAP execution in volatile, dynamically evolving markets by introducing a dynamic neural VWAP framework that integrates a learnable base volume curve, a recurrent adaptive module, and a sequential volume allocation mechanism. By employing either LSTM or TKAN for temporal processing and per-step volume adjustments, the model continually adapts to market feedback to minimize VWAP slippage. Empirical results across five major cryptocurrencies show 10–15% improvements in liquid markets over static neural VWAP baselines, with TKAN-based dynamic models delivering the strongest performance, albeit with higher computational cost. The approach supports real-time deployment and offers flexible extensions, including explicit market-impact modeling and cross-asset dynamics, making it practically relevant for institutional execution in crypto markets.

Abstract

The execution of Volume Weighted Average Price (VWAP) orders remains a critical challenge in modern financial markets, particularly as trading volumes and market complexity continue to increase. In my previous work arXiv:2502.13722, I introduced a novel deep learning approach that demonstrated significant improvements over traditional VWAP execution methods by directly optimizing the execution problem rather than relying on volume curve predictions. However, that model was static because it employed the fully linear approach described in arXiv:2410.21448, which is not designed for dynamic adjustment. This paper extends that foundation by developing a dynamic neural VWAP framework that adapts to evolving market conditions in real time. We introduce two key innovations: first, the integration of recurrent neural networks to capture complex temporal dependencies in market dynamics, and second, a sophisticated dynamic adjustment mechanism that continuously optimizes execution decisions based on market feedback. The empirical analysis, conducted across five major cryptocurrency markets, demonstrates that this dynamic approach achieves substantial improvements over both traditional methods and our previous static implementation, with execution performance gains of 10 to 15% in liquid markets and consistent outperformance across varying conditions. These results suggest that adaptive neural architectures can effectively address the challenges of modern VWAP execution while maintaining computational efficiency suitable for practical deployment.

Recurrent Neural Networks for Dynamic VWAP Execution: Adaptive Trading Strategies with Temporal Kolmogorov-Arnold Networks

TL;DR

This work tackles VWAP execution in volatile, dynamically evolving markets by introducing a dynamic neural VWAP framework that integrates a learnable base volume curve, a recurrent adaptive module, and a sequential volume allocation mechanism. By employing either LSTM or TKAN for temporal processing and per-step volume adjustments, the model continually adapts to market feedback to minimize VWAP slippage. Empirical results across five major cryptocurrencies show 10–15% improvements in liquid markets over static neural VWAP baselines, with TKAN-based dynamic models delivering the strongest performance, albeit with higher computational cost. The approach supports real-time deployment and offers flexible extensions, including explicit market-impact modeling and cross-asset dynamics, making it practically relevant for institutional execution in crypto markets.

Abstract

The execution of Volume Weighted Average Price (VWAP) orders remains a critical challenge in modern financial markets, particularly as trading volumes and market complexity continue to increase. In my previous work arXiv:2502.13722, I introduced a novel deep learning approach that demonstrated significant improvements over traditional VWAP execution methods by directly optimizing the execution problem rather than relying on volume curve predictions. However, that model was static because it employed the fully linear approach described in arXiv:2410.21448, which is not designed for dynamic adjustment. This paper extends that foundation by developing a dynamic neural VWAP framework that adapts to evolving market conditions in real time. We introduce two key innovations: first, the integration of recurrent neural networks to capture complex temporal dependencies in market dynamics, and second, a sophisticated dynamic adjustment mechanism that continuously optimizes execution decisions based on market feedback. The empirical analysis, conducted across five major cryptocurrency markets, demonstrates that this dynamic approach achieves substantial improvements over both traditional methods and our previous static implementation, with execution performance gains of 10 to 15% in liquid markets and consistent outperformance across varying conditions. These results suggest that adaptive neural architectures can effectively address the challenges of modern VWAP execution while maintaining computational efficiency suitable for practical deployment.

Paper Structure

This paper contains 39 sections, 28 equations, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. VWAP Definition and Discretization
  3. Classical VWAP Approaches
  4. Dynamic Volume Approaches
  5. The Rise of Deep Learning in Financial Time Series
  6. Deep Learning Approaches to Market Execution
  7. From Static to Dynamic Neural VWAP
  8. Dynamic VWAP Architecture
  9. Model Overview
  10. Base Volume Curve
  11. Dynamic Volume Adjustment
  12. RNN Architecture Selection
  13. Dynamic Volume Adjustment Network
  14. Sequential Volume Allocation
  15. Base Volume Initialization
  16. ...and 24 more sections

Figures (10)

  • Figure 1: Overview of the dynamic VWAP execution architecture.
  • Figure 2: Slippage between approaches on the full out-of-sample set
  • Figure 3: Slippage between approaches on a subsample of the out-of-sample set
  • Figure 4: Difference in absolute slippage versus naive approach on the full out-of-sample set. Negative values indicate improved performance over the naive approach.
  • Figure 5: Difference in absolute slippage versus naive approach on the subsample set. Negative values indicate improved performance over the naive approach.
  • ...and 5 more figures