eThesis - Kauppakorkeakoulun sähköiset gradut

PURPOSE OF THE STUDY:

Over the last few years, adverse selection has been a widely discussed topic in the equity markets as it has traditionally been seen as having a detrimental effect on a trader's performance. Especially, institutional investors have raised concerns of adverse selection from executing with high-frequency trading (HFT) counterparties. In Europe, today's institutional investors have the opportunity to trade away from the displayed markets and execute orders in dark pools, which should lead to reduced information leakage, and thus also to reduced adverse selection. Although the usage of dark pools is primarily driven by the urge to reduce information leakage, the perception has however grown that even dark pools might leak sensitive information, leading to increased adverse selection. The purpose of this study is to investigate whether dark pools do offer institutional investors the possibility to trade orders with less adverse selection compared to lit markets, and if so, to what extent.

DATA AND METHODOLOGY:

The incidence of adverse selection in lit markets and dark pools is investigated by looking at the stock price development around a Finnish asset management company's resting order fills, which are drawn from the company's algorithmic order flow. The unique data have made it possible for me to examine price moves even in the initial milliseconds around each lit and dark fill. Overall, the data set comprises 152,261 lit market and 16,044 dark pool fills, which occurred during continuous trading hours over the ten-month period from August 1, 2012 to May 31, 2013. All fills are in OMX Helsinki 25 stocks. The lit fills are executed in the Helsinki Stock Exchange and two multilateral trading facilities (MTFs). The dark fills are executed at the primary or the EBBO mid-price in 8 different European MTF dark pools.

RESULTS:

On average, resting orders are getting adversely selected on individual fills in both lit markets and dark pools. However, in dark pools there is less adverse selection, which is the case across the venues and fill size categories. Looking at the whole sample and the average price development over the one-minute post-fill time interval, in lit markets, adverse selection costs are around four times higher compared to dark pools. The stock price starting to move down (up), on average, in less than a second after a buy (sell) order gets filled in a lit market or dark pool is a possible indication of the actions of market participants trading at very high frequency.