The Hunt for
LIS Blocks
Gareth Exton, Global Execution and Quantitative Services, Liquidnet
The trading of blocks in the European equity market has never
been more prevalent, with the large in scale market (LIS) now
accounting for 37% of the on exchange dark activity and on
average 1.59bn 1 Euros being traded above LIS every day.
The combination of regulatory pressure to trade in larger size
while in the dark, the adoption of the conditional order type
and the continuing execution quality challenges of trading in
the lit (e.g. small execution size and information leakage) have
resulted in blocks becoming an increasing part of a trader’s
workflow.
How LIS trading sits within a clients Best Execution policy is
however of increasing concern so here we discuss the issues
involved and a possible solution.
So, while execution quality can be fairly easily proved from a
venue or child execution point of view, how do you prove Best
Execution for the overall trade?
The Value of a Block
Why everyone’s on the hunt.
With the hunt for liquidity a continual battle, the allure of
executing multiple days average daily volume (ADV) in a single
trade has obvious appeal. Couple this with the ability to rest
orders conditionally, particularly now as part of an algorithmic
order, and the result is a growing block market.
20
With Best Execution now a major part of any market
participant’s operational consideration however, block trades
need to perform in terms of execution quality. Studies of
post-trade price movement continually show that block
venues, such as Liquidnet, exhibit the lowest occurrence
of price movement immediately after the execution, and
the lowest absolute price movement in bps. For example,
executions occurring on the Liquidnet MTF experienced a
price movement of less than 0.28bps 2 in the 1 second after
execution, the lowest of any venue.
There are two common problems with the way blocks
are currently measured in TCA, 1) Opportunity cost is
not properly attributed due to the conditional nature
of resting in block venues and 2) the arrival time used
to calculate performance can lead to confusing results,
again due to the conditional nature of the interaction.
www.buysideintel.com
Winter/Spring 2020