Order Flow Characteristics in Nordic Stock Exchanges
Eskelinen, Teemu (2015)
2015
Tuotantotalouden koulutusohjelma
Talouden ja rakentamisen tiedekunta - Faculty of Business and Built Environment
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2015-08-12
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201507301496
https://urn.fi/URN:NBN:fi:tty-201507301496
Tiivistelmä
This thesis studied the statistical characteristics of order flow in Nordic Stock Exchanges. These were also reflected against the recent academic literature in theoretical order flow dynamics. The theoretical and empirical aspects of limit order book markets were explored in a short literature review.
The main focus of the study were the equity exchanges in Helsinki, Stockholm and Copenhagen aspects of which were explored in the empirical part of the thesis. The data set used was NASDAQ OMX Nordic TotalView ITCH feed. Order sizes, inter- arrival times, price impacts and relative positions were extracted from this data set for the period June 2010 – May 2013. These factors are important in, for example, calibrating mathematical order book models for the Nordic Stock Exchanges.
The order flow characteristics mainly followed the outlines set in previous studies conducted in exchanges around the world. In Nordic Exchanges, the order flow was divided into a few very high liquidity stocks and rest of lower liquidity, which made the reporting of generalized descriptive constants difficult. The diurnal seasonality was also found to be very significant and was affected by e.g. the daily openings of U.S. exchanges between 14:00 – 15:00 CET. The relative price power law decay was found to hold with exponents α_hel = 1.15, α_sto = 1.18 and α_cph = 1.20.
Additionally to these descriptive statistics, the study tested the assumptions made by the recent mathematical models. The assumptions were discovered to be too simplistic and e.g. the usage of homogeneous Poisson processes to govern the arrival of order book events was problematic in long time spans. Compared to the Cont et al. (2010) arrivals were reported to be clustered in time, with significant short-term autocorrelation. The arrival times showed humps around 20 milliseconds. The usage of Weibull distributions as total arrival rate process and the Order Flow Imbalance as a short term estimator of immediate price impact were shown to be appropriate for high liquidity stocks with R^2 > 0.40.
The main focus of the study were the equity exchanges in Helsinki, Stockholm and Copenhagen aspects of which were explored in the empirical part of the thesis. The data set used was NASDAQ OMX Nordic TotalView ITCH feed. Order sizes, inter- arrival times, price impacts and relative positions were extracted from this data set for the period June 2010 – May 2013. These factors are important in, for example, calibrating mathematical order book models for the Nordic Stock Exchanges.
The order flow characteristics mainly followed the outlines set in previous studies conducted in exchanges around the world. In Nordic Exchanges, the order flow was divided into a few very high liquidity stocks and rest of lower liquidity, which made the reporting of generalized descriptive constants difficult. The diurnal seasonality was also found to be very significant and was affected by e.g. the daily openings of U.S. exchanges between 14:00 – 15:00 CET. The relative price power law decay was found to hold with exponents α_hel = 1.15, α_sto = 1.18 and α_cph = 1.20.
Additionally to these descriptive statistics, the study tested the assumptions made by the recent mathematical models. The assumptions were discovered to be too simplistic and e.g. the usage of homogeneous Poisson processes to govern the arrival of order book events was problematic in long time spans. Compared to the Cont et al. (2010) arrivals were reported to be clustered in time, with significant short-term autocorrelation. The arrival times showed humps around 20 milliseconds. The usage of Weibull distributions as total arrival rate process and the Order Flow Imbalance as a short term estimator of immediate price impact were shown to be appropriate for high liquidity stocks with R^2 > 0.40.