Date Submitted: 04 Sep 2018
This article qualifies for 0.5 CE under the guidelines of the CFA Institute Continuing Education Program.
We encourage CFA Institute members to login to the CE tracking tool to self-document these credits.
Based on the paper “Heterogeneity in How Algorithmic Traders Impact Institutional Trading Costs” by Tālis J. Putniņš and Joseph Barbara, available at https://www.arx.cfa/post/Heterogeneity-in-how-algorithmic-traders-impactinstitutional-trading-costs-4550.html
This paper was recently recognized for excellence by the CFA Institute Asia-Pacific Research Exchange (ARX) at the 7th Annual Financial Research Network (FIRN) Conference. FIRN is a network of finance researchers and PhD students across Australia and New Zealand.
Traversing the dense, tangled underbrush of an otherwise mostly explored section of securities terrain—the impact of automated, computerized trading—two researchers have demonstrated why it doesn’t pay to ignore the nuances of a complicated subject. Literally, it can cost billions to not heed the observations of authors Putniņš and Barbara, whose paper, “Heterogeneity in How Algorithmic Traders Impact Institutional Trading Costs,” is the subject of this ARX Practitioner’s Brief.
The July 2017 paper is a wake-up call for institutional investors who may not be as vigilant as they think they are when it comes to getting best execution on block orders, if only because their defenses might well be focused on the wrong bad actors, that is, high-frequency traders (HFTs). HFTs, argue Putniņš (University of Technology Sydney) and Barbara (Australian Securities and Investments Commission), are unfairly stigmatized and singled out among computer-program–based or algorithmic traders (ATs) for driving up big-block trade implementation costs when in reality, according to an exhaustive study of trading data, their impact is negligible.
In support of their argument, Putniņš and Barbara fully mapped and surveyed an algorithmic trading community comprising both HFTs, who transact a large number of orders at eye-blink speeds, and non-HFTs. In the process, they uncovered a variety of species and motives, some of which are even beneficial to institutions. On the surface, the ground the authors covered would seem cut and dried: grievances about HFTs have been voiced repeatedly, to the point where no one questions who in this narrative wears the black hat and who wears the white.
What the authors sought to understand was whether the complaints against HFTs had merit. Was there more to the story than what generally has seeped into the mainstream media via books such as Michael Lewis’ Flash Boys
WHAT’S THE INVESTMENT ISSUE?
The rise of electronic equity trading venues at the dawn of the 21st century emptied the trading floors, drove down execution costs, and opened the way for technological advancements, such as order-implementation speeds measured in milliseconds, that few could have ever imagined. By the time of the 2010 flash crash, the fundamental manner by which stocks were traded had radically changed. Although a few die-hard specialists were still clinging to their Big Board posts back on that spring day in 2010, the flash crash made it abundantly clear that algorithms had taken over. At the center of regulatory scrutiny post-flash crash was high-frequency trading, the best-known and most controversial form of algorithmic trading.
With alpha scarce and trading venues fragmented, fund managers increasingly focused their energy on improving execution costs. For decades, the buy side railed against specialists front-running their institutional orders. Now, institutions face a new predator on their blocks: HFTs. These automated strategies account for more than half of the total volume during any given session, and some institutional investors claim they impede liquidity.
As a result of concerns about being preyed upon, institutional investors are forced to break large orders into smaller pieces that need to be traded across multiple venues, making them more susceptible to HFTs. In turn, new liquidity pools and networks have been created to provide a safe space. Yet, as Putniņš and Barbara point out, some studies show that, at best, high-frequency trading and algorithmic trading lower spreads and improve price discovery, and at worst, represented a benign force. So are HFTs good, bad, benign, or what?
HOW DO THE AUTHORS TACKLE THIS ISSUE?
Putniņš and Barbara created a data cross-section reenacting trading of the largest 200 Australian equities (ASX 200 Index constituents) over a 13-month period (1 September 2014 through 30 September 2015), amounting to 273 trading days.
Using unique trader-identified regulatory audit-trail data, they identified a subset of 187 of the most active nondirectional traders (AT/HFT) and measured their activity (roughly 25% of Australian volume on any given day) in terms of the impact on the execution costs for institutions, which control about 80% of Australian large-cap stocks. “Origin of order” identifiers, collected by the Australian Securities and Investments Commission, allowed the authors to reassemble smaller (child) orders back into larger (parent) ones.
Upon close inspection, the AT/HFT gang of 187 proved decidedly heterogeneous. Putniņš and Barbara categorized these traders across a spectrum, ranging from those who drove costs up the highest (toxic) to those who lowered them the most (beneficial).
WHAT ARE THE FINDINGS?
The 12 most toxic traders increased the average order-implementation shortfall cost by 10 basis points or nearly double the cost without the harmful behavior. At the same time, the 14 most beneficial traders systematically decreased costs, effectively, in aggregate, countering the negative impact. However, this offset in aggregate would not have come as any consolation to those individual buyers and sellers specifically impacted by the toxic traders. “An investor that disproportionately interacts with harmful AT/HFT faced higher costs,” concluded the authors.
Interestingly, HFTs were no more likely to be toxic than non-HFTs. And even those ATs/HFTs who drove up costs may have done so unintentionally, merely by trading on the most common entry and exit signals, behavior that could be described not so much as exploitative as lemming-like.
WHAT ARE THE IMPLICATIONS FOR INVESTORS AND INVESTMENT PROFESSIONALS?
First, for buy-side asset managers, it bears underscoring that execution matters. Potentially large cost savings can be realized from trading in a manner that avoids overexposure to toxic counterparties. Such savings could mean the difference between a fund that performs well and one that underperforms.
Second, in terms of execution strategy, more caution should be exercised in smaller stocks, where toxic traders tend to be more active.
Third, effort spent avoiding HFTs may be in vain because many HFTs are beneficial and can reduce institutional execution costs. At the same time, toxic non-HFTs should be avoided if one wants to minimize execution costs.
Finally, from a regulatory perspective, the empirical measurement tools featured in this research could be used to better monitor markets and identify predatory trading behavior.
Summarized by Rich Blake. Rich is a veteran financial journalist who has written for numerous media outlets, including Reuters, ABC News and Institutional Investor. The views expressed herein reflect those of the authors and do not represent the official views of CFA Institute or the authors’ employers.