Traders have long associated adverse selection in dark pools with block executions. A trader typically
considers himself âadversely selectedâ when the stock moves in his favor immediately after he executes
a block in a dark pool. This historical incidence of adverse selection was easy to measure and
detect; the trader simply compared his execution price to the closing price or the price a few hours
later. Moreover, adverse selection in dark pools did not occur systematically in the past; over the long
run, high-performing trades often washed away poorly-performing ones.
Today, however, our research demonstrates that adverse selection occurs systematically in many
dark pools and the nature of it is very different. It transpires over much shorter time periodsâ
over seconds or minutes rather than hours. As such, it may not attract a traderâs attention, but the
accumulating effect of small adverse fills over the longer-term performance of a large parent order
can be considerable.
Todayâs adverse selection is a result of the shift in dark pool composition. Pools that once excluded
high-frequency trading participants have now opened their gates and as a result, experienced
explosive volume growth. Although high-frequency trading firms play an important role in displayed
markets by tightening the spreads, they are often the cause of short-term adverse selection in dark
pools. And, due to the overwhelming participation level of high-frequency trading firms in dark
pools, adverse selection is occurring much more frequently to the detriment of buyside participants.
In this paper, our goal is to educate buyside dark pool participants on the negative effects adverse
selection can have on performance and to provide an effective method for measuring it over the
short-term and long-term. We begin by demonstrating the direct cause and effect relationship between
adverse selection and Implementation Shortfall. We then present our own extended Implementation
Shortfall measurement framework. This method is suitable for measuring the performance
of dark pools as well as dark pool aggregators and it captures the effects of both short-term
and long-term adverse selection. Most importantly, we end with a discussion of various liquidity
filtration techniques that can help traders avoid adverse selection in dark pools.
Understanding and Avoiding
Adverse Selection in Dark Pools
November 2009
©
http://www.itg.com/news_events/papers/AdverseSelectionDarkPools_113009F.pdf
considers himself âadversely selectedâ when the stock moves in his favor immediately after he executes
a block in a dark pool. This historical incidence of adverse selection was easy to measure and
detect; the trader simply compared his execution price to the closing price or the price a few hours
later. Moreover, adverse selection in dark pools did not occur systematically in the past; over the long
run, high-performing trades often washed away poorly-performing ones.
Today, however, our research demonstrates that adverse selection occurs systematically in many
dark pools and the nature of it is very different. It transpires over much shorter time periodsâ
over seconds or minutes rather than hours. As such, it may not attract a traderâs attention, but the
accumulating effect of small adverse fills over the longer-term performance of a large parent order
can be considerable.
Todayâs adverse selection is a result of the shift in dark pool composition. Pools that once excluded
high-frequency trading participants have now opened their gates and as a result, experienced
explosive volume growth. Although high-frequency trading firms play an important role in displayed
markets by tightening the spreads, they are often the cause of short-term adverse selection in dark
pools. And, due to the overwhelming participation level of high-frequency trading firms in dark
pools, adverse selection is occurring much more frequently to the detriment of buyside participants.
In this paper, our goal is to educate buyside dark pool participants on the negative effects adverse
selection can have on performance and to provide an effective method for measuring it over the
short-term and long-term. We begin by demonstrating the direct cause and effect relationship between
adverse selection and Implementation Shortfall. We then present our own extended Implementation
Shortfall measurement framework. This method is suitable for measuring the performance
of dark pools as well as dark pool aggregators and it captures the effects of both short-term
and long-term adverse selection. Most importantly, we end with a discussion of various liquidity
filtration techniques that can help traders avoid adverse selection in dark pools.
Understanding and Avoiding
Adverse Selection in Dark Pools
November 2009
©
http://www.itg.com/news_events/papers/AdverseSelectionDarkPools_113009F.pdf
