Paper: arXiv 2509.25211
Authors: Remi Genet, Hugo Inzirillo
Abstract
This paper introduces Large Execution Models (LEMs), a novel deep learning framework that extends transformer-based architectures to address complex execution problems with flexible time boundaries and multiple execution constraints. Building upon recent advances in neural VWAP execution strategies, LEMs generalize the approach from fixed-duration orders to scenarios where execution duration is bounded between minimum and maximum time horizons, similar to share buyback contract structures. The proposed architecture decouples market information processing from execution allocation decisions: a common feature extraction pipeline using Temporal Kolmogorov-Arnold Networks (TKANs), Variable Selection Networks (VSNs), and multi-head attention mechanisms processes market data to create informational context, while independent allocation networks handle the specific execution logic for different scenarios (fixed quantity vs. fixed notional, buy vs. sell orders). This architectural separation enables a unified model to handle diverse execution objectives while leveraging shared market understanding across scenarios. Through comprehensive empirical evaluation on intraday cryptocurrency markets and multi-day equity trading using DOW Jones constituents, we demonstrate that LEMs achieve superior execution performance compared to traditional benchmarks by dynamically optimizing execution paths within flexible time constraints. The unified model architecture enables deployment across different execution scenarios (buy/sell orders, varying duration boundaries, volume/notional targets) through a single framework, providing significant operational advantages over asset-specific approaches.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematics including Temporal Kolmogorov-Arnold Networks (TKANs), multi-head attention, and optimal control formulations, while providing comprehensive empirical evaluation on real-world datasets (cryptocurrency and DOW Jones) with reported performance improvements.
Research Flowchart
flowchart TD A["Research Goal: Generalize VWAP Execution<br>to Flexible Time Constraints"] --> B["Methodology: LEM Architecture Design"] B --> C["Data Inputs:<br>Intraday Crypto & Multi-Day Equity Markets"] C --> D["Core Computation: Market Information Processing<br>TKANs + VSNs + Multi-Head Attention"] D --> E["Parallel Execution Allocation Networks<br>Fixed Qty vs. Fixed Notional | Buy vs. Sell"] E --> F["Key Findings:<br>1. Superior execution vs. traditional benchmarks<br>2. Dynamic optimization in flexible time horizons<br>3. Unified model for multi-scenario deployment"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper introduces Large Execution Models (LEMs), a novel deep learning framework that extends transformer-based architectures to address complex execution problems with flexible time boundaries and multiple execution constraints.
Source sentence
This paper introduces Large Execution Models (LEMs), a novel deep learning framework that extends transformer-based architectures to address complex execution problems with flexible time boundaries and multiple execution constraints.
Assumes 2
- The ability to predict future volumes accurately. model
Source sentence
the ability to predict future volumes accurately
- Statistical independence between prediction errors and price volatility. statistical
Source sentence
the statistical independence between prediction errors and price volatility
Data
- Cryptocurrency Market Dataset market-prices commercial
Source sentence
For the cryptocurrency evaluation, we utilize minutely candlestick data directly sourced from Binance Data Vision, encompassing twenty cryptocurrency spot trading pairs
- Equity Market Dataset market-prices public
Source sentence
For traditional equity markets, we focus on thirty stocks comprising the Dow Jones Industrial Average
Costs
Modeled
Finds 3
- LEMs achieve an average of -35bps for buying and +36bps for selling when given sufficient time flexibility, outperforming the benchmark without substantially increased risk. performance
Source sentence
the LEM achieves an average of -35bps for buying and +36bps for selling when given sufficient time flexibility (such as completing execution in half the allocated time).
- With a 15‑minute frequency, the model yields an average buy slippage 7bps lower than the benchmark and an average sell price nearly 9bps above the benchmark. performance
Source sentence
the lowest time duration constraint presents similar values across the first seven minimum periods, with an average buy slippage 7 basis points lower than the benchmark and average sell price nearly 9 basis points above the benchmark.
- The model reduces execution risk, achieving a 20% risk reduction for the Match VWAP component compared to other strategies. performance
Source sentence
the Match VWAP component outperforms all others in terms of risk, with a 20% reduction in risk.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 22 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (111)
- The Kolmogorov–Arnold representation theorem revisited (2021)
- Semi-supervised classification with graph convolutional networks (2016) — link
- Wav-kan: Wavelet kolmogorov-arnold networks, 2024 — link
- Optimal slice of a VWAP trade (2002) — link
- Optimal VWAP Trading Strategy and Relative Volume (2007) — link
- Non-linear time series: a dynamical system approach (1990)
- Concave Cross Impact (2024) — link
- Measuring the Information Content of Stock Trades (1991)
and 103 more
- Trading Cost: The Critical Link Between Investment Information and Results (1983)
- Optimal Execution Strategies in Limit Order Books with General Shape Functions (2010)
- Fluctuations and Response in Financial Markets: The Subtle Nature of Random Price Changes (2004)
- Temporal fusion transformers for interpretable multi-horizon time series forecasting (2021)
- TimesKAN: A Generic Backbone Network for Time Series Prediction Based on Kolmogorov-Arnold Networks (2024)
- Crossformer: Transformer utilizing cross-dimension dependency for multivariate time series forecasting (2023)
- Integrating Kolmogorov–Arnold Networks with Time Series Prediction Framework in Electricity Demand Forecasting (2025)
- Gkan: Graph kolmogorov-arnold networks (2024) — link
- Deep residual learning for image recognition (2016)
- Informer: Beyond efficient transformer for long sequence time-series forecasting (2021)
- LPSNet: a novel log path signature feature based hand gesture recognition framework (2017)
- Extracting information from the signature of a financial data stream (2013) — link
- Deep signature statistics for likelihood-free time-series models (2021)
- Embedding and learning with signatures (2021)
- Chebyshev polynomial-based kolmogorov-arnold networks: An efficient architecture for nonlinear function approximation (2024) — link
- Smooth Kolmogorov Arnold networks enabling structural knowledge representation (2024) — link
- Graphkan: Enhancing feature extraction with graph kolmogorov arnold networks (2024) — link
- Time series analysis: forecasting and control (2015)
- Enhancing the locality and breaking the memory bottleneck of transformer on time series forecasting (2019)
- Hippo-kan: Efficient kan model for time series analysis (2024) — link
- Tsmixer: An all-mlp architecture for time series forecasting (2023) — link
- KAN: Kolmogorov-Arnold Networks (2024) — link
- Sigmoid-weighted linear units for neural network function approximation in reinforcement learning (2018)
- Vector autoregressive models for multivariate time series (2006)
- Layer Normalization (2016) — link
- Vector autoregressive models (2013)
- State-space models (1994)
- Dynamic linear models with Markov-switching (1994)
- Time series forecasting using neural networks vs. Box-Jenkins methodology (1991)
- Connecting the dots: Multivariate time series forecasting with graph neural networks (2020)
- Multiple convolutional neural networks for multivariate time series prediction (2019)
- Neural networks: a comprehensive foundation (1998)
- Approximation by superpositions of a sigmoidal function (1989)
- Autoregressive convolutional neural networks for asynchronous time series (2018)
- A novel LSTM approach for asynchronous multivariate time series prediction (2019)
- Multilayer feedforward networks are universal approximators (1989)
- The vanishing gradient problem during learning recurrent neural nets and problem solutions (1998)
- Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling (2014)
- Learning long-term dependencies with gradient descent is difficult (1994) — link
- Backpropagation through time: what it does and how to do it (1990)
- A learning algorithm for continually running fully recurrent neural netwokrs (1989)
- Long short-term memory (1997)
- On the Properties of Neural Machine Translation: Encoder-Decoder Approaches (2014)
- A simple way to initialize recurrent networks of rectified linear units (2015) — link
- Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs) (2016)
- Multi-horizon time series forecasting with temporal attention learning (2019)
- Financial time series forecasting with deep learning: A systematic literature review: 2005–2019 (2020)
- Study of effectiveness of time series modeling (ARIMA) in forecasting stock prices (2014)
- ARMA models and the Box–Jenkins methodology (1997)
- Recurrent neural networks (2001)
- Recurrent neural networks for time series forecasting: Current status and future directions (2021)
- On the representation of continuous functions of several variables by superpositions of continuous functions of a smaller number of variables (1961)
- The perceptron: A probabilistic model for information storage and organization in the brain. (1958)
- Understanding LSTM–a tutorial into long short-term memory recurrent neural networks (2019) — link
- Learning phrase representations using RNN encoder-decoder for statistical machine translation (2014) — link
- Deep learning for time series forecasting: a survey (2021)
- Time-series forecasting with deep learning: a survey (2021)
- Attention is all you need (2017)
- Fast and accurate deep network learning by exponential linear units (elus) (2015) — link
- Language modeling with gated convolutional networks (2017)
- Kolmogorov-arnold networks for time series: Bridging predictive power and interpretability (2024) — link
- seqKAN: Sequence processing with Kolmogorov-Arnold Networks (2025) — link
- Tkan: Temporal kolmogorov-arnold networks (2024) — link
- C-kan: A new approach for integrating convolutional layers with kolmogorov–arnold networks for time-series forecasting (2024)
- Sigkan: Signature-weighted kolmogorov-arnold networks for time series (2024) — link
- KAN-AD: Time series anomaly detection with Kolmogorov-Arnold networks (2024) — link
- Kolmogorov-arnold networks (kan) for time series classification and robust analysis (2024)
- Kolmogorov-arnold networks (kans) for time series analysis (2024) — link
- Abnormality Detection in Time-Series Bio-Signals using Kolmogorov-Arnold Networks for Resource-Constrained Devices (2024)
- Kolmogorov–Arnold recurrent network for short term load forecasting across diverse consumers (2025)
- TSKANMixer: Kolmogorov-Arnold Networks with MLP-Mixer Model for Time Series Forecasting (2025) — link
- Kolmogorov-Arnold Networks for Time Series Granger Causality Inference (2025) — link
- An algorithm for the machine calculation of complex Fourier series (1965)
- Zero Shot Time Series Forecasting Using Kolmogorov Arnold Networks (2024) — link
- Exploring kolmogorov-arnold networks for interpretable time series classification (2024) — link
- Integration of paths–A faithful representation of paths by noncommutative formal power series (1958)
- Rough paths, signatures and the modelling of functions on streams (2014) — link
- Differential equations driven by rough signals (1998)
- A primer on the signature method in machine learning (2016) — link
- Chebyshev polynomials (2002)
- Hippo: Recurrent memory with optimal polynomial projections (2020)
- Developing the path signature methodology and its application to landmark-based human action recognition (2022)
- N-BEATS: Neural basis expansion analysis for interpretable time series forecasting (2019) — link
- Domain-adversarial training of neural networks (2016)
- A CNN–LSTM model for gold price time-series forecasting (2020)
- Shape and time distortion loss for training deep time series forecasting models (2019)
- Recurrent Fourier-Kolmogorov Arnold Networks for photovoltaic power forecasting (2025)
- TimeKAN: KAN-based Frequency Decomposition Learning Architecture for Long-term Time Series Forecasting (2025) — link
- A gated residual kolmogorov-arnold networks for mixtures of experts (2024) — link
- Dispensing with optimal control: a new approach for the pricing and management of share buyback contracts — → 2404.13754
- A new method for the pricing of share buy-back programmes (2024)
- Optimal execution of accelerated share repurchase contracts with fixed notional (2017)
- Accelerated share repurchase: pricing and execution strategy (2015)
- Optimal accelerated share repurchase (2017)
- Valuing american options by simulation: a simple least-squares approach (2001)
- Deep Learning for VWAP Execution in Crypto Markets: Beyond the Volume Curve — → 2502.13722
- VWAP Execution with Signature-Enhanced Transformers: A Multi-Asset Learning Approach — → 2503.02680
- Recurrent neural networks for dynamic vwap execution: Adaptive trading strategies with temporal kolmogorov-arnold networks — → 2502.18177
- A temporal kolmogorov-arnold transformer for time series forecasting (2024) — link
- Sigkan: Signature-weighted kolmogorov-arnold networks for time series (2024) — link
- Optimal slice of a vwap trade (2002)
- Optimal VWAP trading under noisy conditions (2011)
- Improving VWAP strategies: A dynamic volume approach (2008)
4 of 111 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.