Search
Close this search box.
Search
Close this search box.

Why do traders split orders?

Why do traders split orders?

Trading and Investing

The research work “Quantitative statistical analysis of order-splitting behaviour of individual trading accounts in the Japanese stock market over nine years” by Yuki Sato and Kiyoshi Kanazawa presents a comprehensive analysis of order-splitting behavior in trading, focusing on the Japanese stock market.

Furthermore, Rebellion Research would like to thank Capital Fund Management’s Jean-Philippe Bouchaud for bringing this excellent piece of research to our attention!

Tokyo Stock Exchange (1950)
More details. Tokyo Stock Exchange (1950)

Now let’s peruse the key aspects, methodologies, and implications of this work!

Overview

The paper investigates the order-splitting strategy, wherein large potential metaorders are executed in smaller segments to minimize transaction costs. This strategy is crucial in understanding the long-range correlation (LRC) in persistent order flow, a phenomenon observed in stock market trading.

Background and Methodology

  1. Theoretical Basis: The study builds upon the model introduced by Lillo, Mike, and Farmer (LMF) in 2005, which proposed a microscopic model of order-splitting traders. This model aimed to quantitatively predict the LRC’s behavior from microscopic dynamics.
  2. Data Analysis: Sato and Kanazawa analyzed a substantial dataset from the Tokyo Stock Exchange (TSE) covering nine years. This dataset included account data of traders, referred to as virtual servers. Through meticulous preprocessing, the authors were able to effectively categorize trader IDs.
  3. Strategy Clustering: The researchers applied strategy clustering to individual traders, distinguishing between order-splitting traders and random traders.

Key Findings

Old Tokyo Stock Exchange building (c. 1960)
More details. Old Tokyo Stock Exchange building (c. 1960)
  1. Metaorder Length Distribution: The study found that for most stocks, the metaorder length distribution follows power laws with an exponent α, conforming to the formula P(L)∝L−α−1, with L representing the metaorder length.
  2. Verification of LMF Prediction: By analyzing sign correlation C(τ)∝τ−γ, the research directly confirmed the LMF model’s prediction that γ≈α−1.
  3. Estimation of Splitting Traders: The paper discusses a methodology to estimate the total number of splitting traders using publicly available data, based on the Autocorrelation Function (ACF) prefactor formula in the LMF model.

Implications and Significance

This research is significant as it provides the first quantitative evidence supporting the LMF model, particularly in the context of the Japanese stock market. The findings have several implications:

  1. Understanding Market Dynamics: The research enhances understanding of how order-splitting behavior impacts market dynamics, particularly in terms of long-range correlations in order flows.
  2. Model Validation: The study validates the LMF model quantitatively, offering a solid foundation for future research in this area.
  3. Trading Strategies: Insights from the study can inform trading strategies and risk management approaches, especially for large institutional traders.
  4. Regulatory Implications: Understanding order-splitting behavior can also have regulatory implications, helping to monitor and manage market manipulation strategies.

The paper by Sato and Kanazawa represents an interesting quantitative analysis of order splitting. By leveraging extensive data from the Tokyo Stock Exchange and applying rigorous statistical methods, the researchers have provided valuable insights into the microdynamics of stock markets. Lastly, their work not only validates existing theoretical models but also opens avenues for further research in financial market dynamics.

Read the full paper: 2308.01112.pdf (arxiv.org)

Why do traders split orders?