Forecasting Intraday Volume in Equity Markets with Machine Learning
This study employs machine learning, clustering, and DeepLOBv to forecast intraday trading volumes and improve VWAP execution strategies.
What it examines
This study develops machine learning models to forecast intraday trading volume, enhancing the classical CMEM with high-frequency predictors and clustering across stocks. Using schemes like SAM, CAM, and UAM along with models such as DeepLOBv, it aims to improve portfolio execution and VWAP strategy performance.
What it concludes
The results demonstrate improved intraday volume prediction, leading to better VWAP replication and trading execution. Applications include reducing market impact for large orders and enhancing order fill rates. Future research should integrate more market features and volatility to further refine portfolio execution strategies.
Evidence objects
Researchers demonstrate that machine learning techniques, notably deep neural networks with modified DeepLOB architecture integrating CNN and LSTM layers, outperform traditional models, improving intraday trading volume predictability in equity markets.
key_findings bullet 1 · key_findings · validation V0
The study introduces innovative clustered and universal asset models to incorporate crosssectional commonality, while incorporating auxiliary features and order flow imbalance techniques, unveiling enhancements to liquidity dynamics and CMEM performance.
key_findings bullet 2 · key_findings · validation V0
Utilizing highfrequency limit order book data, dynamic prediction settings, and regularization methods like Ridge and LASSO, researchers develop compound predictors, though realtime adaptability in volatile conditions remains an unresolved challenge.
key_findings bullet 3 · key_findings · validation V0
The paper introduces a novel machine learning framework for forecasting intraday trading volumes in equity markets by integrating high-frequency predictors with clustering methods. Its originality lies in leveraging asset commonality, offering a flexible alternative to traditional additive models, making it a compelling and unique addition to modern high-frequency trading research.
key_findings bullet 4 · key_findings · validation V0
Raw abstract and provenance
Abstract: This study focuses on forecasting intraday trading volumes, a crucial component for portfolio implementation, especially in high-frequency (HF) trading environments. Given the current scarcity of flexible methods in this area, we employ a suite of machine learning (ML) models enriched with numerous HF predictors to enhance the predictability of intraday trading volumes. Our findings reveal that in… ▽ More This study focuses on forecasting intraday trading volumes, a crucial component for portfolio implementation, especially in high-frequency (HF) trading environments. Given the current scarcity of flexible methods in this area, we employ a suite of machine learning (ML) models enriched with numerous HF predictors to enhance the predictability of intraday trading volumes. Our findings reveal that intraday stock trading volume is highly predictable, especially with ML and considering commonality. Additionally, we assess the economic benefits of accurate volume forecasting through Volume Weighted Average Price (VWAP) strategies. The results demonstrate that precise intraday forecasting offers substantial advantages, providing valuable insights for traders to optimize their strategies. △ Less
Source row: 901 · abstract type: unknown