Affiliations:
School of Business Management, NMIMS University, Mumbai, India
Correspondence:
[*]
Corresponding author: Sudhanshu Pani, School of Business Management, NMIMS University, Mumbai - 400 056, India. Tel: +91-022-42355886, E-mail: sudhanshu.pani@sbm.nmims.edu.
Abstract: The pattern of dependence between liquidity, durations (orders and trades) and bid-ask spreads in a limit order market are examined in high resolution invoking copulas and graph theory. Using intraday data from a sample of NASDAQ 100 stocks and an experimental design, we study the information pathways in markets in the presence of algorithmic traders. Our results confirm that multivariate analysis is more appropriate to investigate these information pathways. We observe that the strength and nature of the dependence between variables vary through the trading day. We confirm the existence of stylised aspects of algorithmic trading, such as tail dependence in trade durations, a balance between buy and sell side in order durations, liquidity and bid-ask spreads, and the bid-ask spread and liquidity trade-off in the dependence structure.
Keywords: High resolution, Limit order markets, Durations, Liquidity, Bid-Ask Spread, Algorithmic trading, Pair copula construction
How to investigate Information pathways in limit order markets, when viewed in high resolution?
What is it about?
Modern markets have a large number of algorithmic traders are participants. This paper captures stylised aspects of their behaviour using public datasets from an exchange. When limit order markets are viewed in high resolution or ultra high frequency, the dependence between the market microstructure variables can help explain information flow in market. The information flow relates to how the traders bring information into the market. This could be through the price (spread), quantity or volume of shares in their quotes, arrival rates into the market or time of arrival, transactions. This paper describes the methodology of how the dependence between market microstructure variables can be investigated and used by market participants.
Why is it important?
This research adds to the market microstructure toolbox. The methods currently used in market microstructure are not best suited to be employed in high resolution or ultra high frequency empirical investigation. VAR, ACD etc based methods are suitable for coarser analysis as they are designed under implicit assumption of equilibrium markets.