Paper: arXiv 2503.02680

Abstract

In this paper I propose a novel approach to Volume Weighted Average Price (VWAP) execution that addresses two key practical challenges: the need for asset-specific model training and the capture of complex temporal dependencies. Building upon my recent work in dynamic VWAP execution arXiv:2502.18177, I demonstrate that a single neural network trained across multiple assets can achieve performance comparable to or better than traditional asset-specific models. The proposed architecture combines a transformer-based design inspired by arXiv:2406.02486 with path signatures for capturing geometric features of price-volume trajectories, as in arXiv:2406.17890. The empirical analysis, conducted on hourly cryptocurrency trading data from 80 trading pairs, shows that the globally-fitted model with signature features (GFT-Sig) achieves superior performance in both absolute and quadratic VWAP loss metrics compared to asset-specific approaches. Notably, these improvements persist for out-of-sample assets, demonstrating the model’s ability to generalize across different market conditions. The results suggest that combining global parameter sharing with signature-based feature extraction provides a scalable and robust approach to VWAP execution, offering significant practical advantages over traditional asset-specific implementations.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper uses advanced mathematical concepts like path signatures and transformer architectures, indicating high complexity. It also provides strong empirical evidence with backtests on 80 cryptocurrency pairs, comparing performance metrics, which suggests high implementation readiness.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Scalable VWAP Execution"] --> B["Data Collection:<br>80 Crypto Trading Pairs"]
  B --> C["Methodology:<br>GFT-Sig Transformer"]
  C --> D{"Training Strategy"}
  D --> E["Global Model<br>Multi-Asset Learning"]
  D --> F["Baseline<br>Asset-Specific Models"]
  E & F --> G["Evaluation:<br>VWAP Loss Metrics"]
  G --> H["Outcome:<br>GFT-Sig Outperforms Baselines<br>+ Cross-Asset Generalization"]

Cited References (127)

  1. High frequency trading under scrutiny (2009)
  2. The Total Cost of Transactions on the NYSE (1988)
  3. VWAP strategies, Transaction Performance: The Changing Face of Trading Investment Guides Series (2002)
  4. Optimal execution of portfolio transactions (2000)
  5. Optimal slice of a VWAP trade (2002) — link
  6. Optimal VWAP Trading Strategy and Relative Volume (2007) — link
  7. Mean Variance Optimal VWAP Trading (2012) — link
  8. Price, trade size, and information in securities markets (1987) — link
and 119 more
  1. A Theory of the Interday Variations in Volume, Variance, and Trading Costs in Securities Markets (1990) — link
  2. The Price Variability-Volume Relationship on Speculative Markets (1983) — link
  3. The Relation between Price Changes and Trading Volume: A Survey (1987)
  4. Intra-day market activity (1999)
  5. Optimal VWAP trading under noisy conditions (2011)
  6. Optimal execution of a vwap order: A stochastic control approach (2013) — link
  7. Dynamic Programming and Trade Execution (2013)
  8. VWAP Execution and Guaranteed VWAP (2014)
  9. Are the Trading Volume and the Number of Trades Distributions Universal (2011)
  10. Permanent Market Impact Can Be Nonlinear (2013)
  11. Hierarchical Deep Reinforcement Learning for VWAP Strategy Optimization — → 2212.14670
  12. An Adaptive Dual-level Reinforcement Learning Approach for Optimal Trade Execution — → 2307.10649
  13. An Optimal Control Strategy for Execution of Large Stock Orders Using LSTMs — → 2301.09705
  14. TKAN: Temporal Kolmogorov-Arnold Networks (2024) — link
  15. SigKAN: Signature-Weighted Kolmogorov-Arnold Networks for Time Series (2024) — link
  16. A Temporal Kolmogorov-Arnold Transformer for Time Series Forecasting (2024) — link
  17. A Gated Residual Kolmogorov-Arnold Networks for Mixtures of Experts (2024)
  18. A Temporal Linear Network for Time Series Forecasting (2024)
  19. Are Transformers Effective for Time Series Forecasting? (2022) — link
  20. TSMixer: An All-MLP Architecture for Time Series Forecasting (2023) — link
  21. Long short-term memory (1997)
  22. Learning phrase representations using RNN encoder-decoder for statistical machine translation (2014)
  23. Neural machine translation by jointly learning to align and translate (2014) — link
  24. Attention is all you need (2017)
  25. Deep Smoothing of the Implied Volatility Surface (2020)
  26. Deep Learning Volatility: A deep neural network perspective on pricing and calibration in (rough) volatility models (2019) — link
  27. Financial time series forecasting with deep learning: A systematic literature review: 2005–2019 (2020)
  28. Deep learning models for price forecasting of financial time series: A review of recent advancements: 2020-2022 — → 2305.04811
  29. Integration of paths–A faithful representation of paths by noncommutative formal power series (1958)
  30. Rough paths, signatures and the modelling of functions on streams (2014) — link
  31. Differential equations driven by rough signals (1998)
  32. A primer on the signature method in machine learning (2016) — link
  33. Developing the path signature methodology and its application to landmark-based human action recognition (2022)
  34. LPSNet: a novel log path signature feature based hand gesture recognition framework (2017)
  35. Extracting information from the signature of a financial data stream (2013) — link
  36. Deep signature statistics for likelihood-free time-series models (2021)
  37. Embedding and learning with signatures (2021)
  38. A signature-based machine learning model for distinguishing bipolar disorder and borderline personality disorder (2018)
  39. Multilayer feedforward networks are universal approximators (1989)
  40. The perceptron: A probabilistic model for information storage and organization in the brain. (1958)
  41. Approximation by superpositions of a sigmoidal function (1989)
  42. On the representation of continuous functions of several variables by superpositions of continuous functions of a smaller number of variables (1961)
  43. KAN: Kolmogorov-Arnold Networks (2024) — link
  44. Multiple-output modeling for multi-step-ahead time series forecasting (2010)
  45. Model specification in multivariate time series (1989)
  46. Multivariate time series analysis and applications (2018)
  47. A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series (2006)
  48. Time series forecasting using neural networks vs. Box-Jenkins methodology (1991)
  49. State-space models (1994)
  50. Dynamic linear models with Markov-switching (1994)
  51. Vector autoregressive models for multivariate time series (2006)
  52. Vector autoregressive models (2013)
  53. Autoregressive convolutional neural networks for asynchronous time series (2018)
  54. A novel LSTM approach for asynchronous multivariate time series prediction (2019)
  55. Fast and accurate deep network learning by exponential linear units (elus) (2015) — link
  56. Language modeling with gated convolutional networks (2017)
  57. Layer Normalization (2016) — link
  58. Long short-term memory (1997)
  59. Wav-KAN: Wavelet Kolmogorov-Arnold Networks (2024) — link
  60. Kolmogorov-arnold networks (kans) for time series analysis (2024) — link
  61. Temporal fusion transformers for interpretable multi-horizon time series forecasting (2021)
  62. Sigmoid-weighted linear units for neural network function approximation in reinforcement learning (2018)
  63. Twenty years of mixture of experts (2012)
  64. Temporal pattern attention for multivariate time series forecasting (2019)
  65. Multivariate temporal convolutional network: A deep neural networks approach for multivariate time series forecasting (2019)
  66. A survey of transformers (2022)
  67. Delight: Deep and light-weight transformer (2020) — link
  68. Exploring the limits of transfer learning with a unified text-to-text transformer (2020)
  69. The evolved transformer (2019)
  70. Mask attention networks: Rethinking and strengthen transformer (2021) — link
  71. Funnel-transformer: Filtering out sequential redundancy for efficient language processing (2020)
  72. Attentive Language Models Beyond a Fixed-Length Context. Published online June 2, 2019. doi: 10.48550 (1901)
  73. Megatron-lm: Training multi-billion parameter language models using model parallelism (2019) — link
  74. Efficient content-based sparse attention with routing transformers (2021)
  75. Compressive transformers for long-range sequence modelling (2019) — link
  76. TENER: adapting transformer encoder for named entity recognition (2019) — link
  77. Tuning multilingual transformers for language-specific named entity recognition (2019)
  78. Improving multimodal named entity recognition via entity span detection with unified multimodal transformer (2020)
  79. Learning spatio-temporal transformer for visual tracking (2021)
  80. Spatio-temporal transformer network for video restoration (2018)
  81. Generating long sequences with sparse transformers (2019) — link
  82. Position-aware attention and supervised data improve slot filling (2017)
  83. Bp-transformer: Modelling long-range context via binary partitioning (2019) — link
  84. Generating wikipedia by summarizing long sequences (2018) — link
  85. Linformer: Self-attention with linear complexity (2020) — link
  86. Poolingformer: Long document modeling with pooling attention (2021)
  87. Informer: Beyond efficient transformer for long sequence time-series forecasting (2021)
  88. Deep transformer models for time series forecasting: The influenza prevalence case (2020) — link
  89. Are transformers effective for time series forecasting? (2023)
  90. Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting (2019)
  91. Adversarial sparse transformer for time series forecasting (2020)
  92. Understanding LSTM–a tutorial into long short-term memory recurrent neural networks (2019) — link
  93. A survey on vision transformer (2022)
  94. Connecting the dots: Multivariate time series forecasting with graph neural networks (2020)
  95. Multiple convolutional neural networks for multivariate time series prediction (2019)
  96. Adaptive mixtures of local experts (1991)
  97. DeepAR: Probabilistic forecasting with autoregressive recurrent networks (2020)
  98. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019) — link
  99. Deep state space models for time series forecasting (2018)
  100. Hierarchical mixtures of experts and the EM algorithm (1994)
  101. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer (2017) — link
  102. Network of experts for large-scale image categorization (2016)
  103. Expert gate: Lifelong learning with a network of experts (2017)
  104. Scaling vision with sparse mixture of experts (2021)
  105. A multilevel mixture-of-experts framework for pedestrian classification (2011)
  106. Mixture of experts: a literature survey (2014)
  107. Glam: Efficient scaling of language models with mixture-of-experts (2022)
  108. Kolmogorov-Arnold Networks for Time Series: Bridging Predictive Power and Interpretability (2024) — link
  109. A Temporal Kolmogorov-Arnold Transformer for Time Series Forecasting (2024) — link
  110. Learning interpretable deep state space model for probabilistic time series forecasting (2021) — link
  111. Deep State Space Recurrent Neural Networks for Time Series Forecasting (2024) — link
  112. Mixture of MLP-experts for trend forecasting of time series: A case study of the Tehran stock exchange (2011)
  113. Autoformer: Decomposition transformers with auto-correlation for long-term series forecasting (2021)
  114. Pyraformer: Low-complexity pyramidal attention for long-range time series modeling and forecasting (2021)
  115. FEDformer: Frequency enhanced decomposed transformer for long-term series forecasting (2022)
  116. SigGate: Enhancing Recurrent Neural Networks with Signature-Based Gating Mechanisms (2025) — link
  117. Deep Learning for VWAP Execution in Crypto Markets: Beyond the Volume Curve — → 2502.13722
  118. Keras Sig: Efficient Path Signature Computation on GPU in Keras 3 (2025) — link
  119. Recurrent Neural Networks for Dynamic VWAP Execution: Adaptive Trading Strategies with Temporal Kolmogorov-Arnold Networks — → 2502.18177

6 of 127 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.