Machine learning models distinguish high-frequency trading strategies using market data

Researchers have developed a novel method to distinguish between liquidity-supplying and liquidity-demanding high-frequency trading (HFT) strategies using machine learning models trained on proprietary Nasdaq data. This approach generates daily HFT measures for U.S. stocks from 2010 to 2023, which prove more effective than standard proxies and capture previously missed time-series variations. The methodology has been validated using Euronext Paris data, demonstrating its cross-market generalizability and predictive power. Analysis of the 14-year dataset indicates that supply-side HFT is linked to enhanced pre-announcement information acquisition and reduced bid-ask spreads, while demand-side HFT shows opposite correlations. Notably, during the COVID-19 pandemic, HFT-supplied liquidity remained robust and its association with lower spreads intensified. AI

IMPACT Enhances understanding of market dynamics and HFT strategies, potentially informing regulatory approaches and trading algorithms.

RANK_REASON Academic paper detailing a new methodology for analyzing financial market data using machine learning. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• Ben Moews
• COVID-19
• Euronext Paris
• Nasdaq

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →