The crux of this question is whether there is anything new
or particular to high-frequency trading (HFT) that demands
special regulatory attention. Definitions of HFT vary, and the
label is imprecise as it groups together a variety of distinct
practices and tactics that may affect markets in different
ways. Nevertheless, all agree that HFT is characterised by
extremely high speed, and autonomous operation. Both of these
properties can cause markets to perform in modes not possible
before the advent of HFT, and we have substantial evidence of
anomalies caused by HFT activity. So what is it that makes HFT
qualitatively different, such that it needs regulatory
First, the very speed at which HFT operates introduces
qualitatively new patterns of activity. High-frequency traders
receive and process information from exchanges with ultra-low
latency measured in milliseconds, and are able to react with
orders within a few milliseconds more. Such speed provides an
edge in responding to economic news, which translates to
profitable advantage as the first responder reaps the benefit
of executing against mispriced outstanding orders.
substantial evidence of anomalies caused by HFT
Most of the time, however, there is little substantial news
– in the majority of milliseconds, nothing of
real-world economic consequence actually happens. As such, the
advantage is actually deployed in response to events within the
Momentary mispricing arises even in the absence of news,
simply due to the short-term imbalances of supply and demand
induced by order flow. The winner-takes-all nature of response
to mispricing leads inevitably to a latency arms race.
This sees competing high-frequency traders take
extraordinary measures involving enormous investments in
technology and communications infrastructure to shave
microseconds off their response time and thereby improve their
odds of winning the profit opportunities that arise throughout
the trading day.
Second, the autonomy of HFT increases the difficulty of
anticipating its effects in complex or unusual situations.
Autonomous operation is of course necessary to compete in the
latency arms race, and the ability to process high volumes of
information at great speed is a great spur to automation.
Human traders may likewise be unpredictable, but we have
hundreds of years of experience with them, and the relative
slowness of human-scale trading can limit damage accrued before
measures are taken.
That HFT may introduce different behaviours to the market is
not in itself a bad thing. Indeed, the general progress in
automation of trading is quite likely responsible for
reductions in transaction costs and improvement in price
discovery. What the point about qualitative differences does
entail a priori is that HFT deserves special scrutiny
in case such extremes of speed and autonomy contribute to
systemic risks, or allow particular trading tactics that
detract from efficiency or otherwise degrade the performance of
The a priori argument is only strengthened by
evidence from experience. The most notorious market HFT anomaly
to date was the so-called Flash Crash of May 2010. Whereas some
consider HFT to have been exonerated as the proximal cause of
that incident, many knowledgeable observers argue that the
predominance of HFT actors in the affected markets enabled the
subsequent response, by suddenly withdrawing the liquidity they
had generally been providing. And although the Flash Crash left
no permanent damage and has not been repeated, as far as we
understand it could happen again at any time. In fact, it has
been reported that so-called mini flash crashes in individual
securities have actually been occurring with some regularity in
The question of flash crashes underscores how little we
actually understand about the implications of pervasive HFT
activity. Numerous academic studies on HFT in the last few
years – based on theoretical models, or empirical data
analysis – have shed light on aspects of HFT. But the
overall picture remains incomplete at best.
Standard theoretical finance models, generally speaking,
abridge the fine granularity of information transmission time
scales driving the latency arms race. And even the most
thorough data analyses have difficulty identifying the effects
of particular HFT strategies. What we care most about
– the potential exacerbation of rare and extreme
market events – is inherently difficult to study.
This is because, by definition, the conditions present
themselves only rarely.
All this suggests a need for more fundamental research for
the long-term, and continued vigilance in the short-term.
Prudence in the face of uncertainty calls for regulatory
scrutiny, but also dictates a posture of humility and
cautiousness in promulgating specific measures.
The most effective lever
The proper goal of regulatory oversight is not to diminish
automated trading per se, but rather to promote the
beneficial HFT tactics and discourage the practices deemed
Prohibiting the perceived harmful tactics directly is often
difficult to enforce, given that trading strategies are
In many cases, the most effective lever is the design of
market rules that align incentives, by rendering the
discouraged strategies more costly or less effective.
By Michael P Wellman, professor of computer science and
engineering at the University of Michigan
Middlemen have never been popular. Something just feels
wrong about paying someone to be introduced to a counterparty
you could theoretically find on your own. Monopolies have never
been popular, either, especially cartels conspiring to jack up
prices. So what should we think about cartels of middlemen
creating monopolies to jack up prices?
We start this argument against regulating HFT with this
question in the hopes that, when we come back to it later, it
will help us better understand why we're in today's dilemma.
And we are in a dilemma. Every retail investor you ask will
tell you that the market is rigged by HFTs, whose speed,
collocation, algorithms, and automated cancellation of
thousands of what are deemed phony orders are causing a gaping
lack of confidence in the Securities and Exchange Commission's
(SEC) all-electronic National Market System (NMS).
But if you run the actual numbers, trading has never been
cheaper or easier for the retail investor. The amount
high-frequency trader's earn, even in good times averaging only
a tenth of a cent per share, is dropping like a stone as more
competitors enter the business. The sector earned $1.25 billion
in 2012, down three quarters since 2009. No conspiring cartels
jacking up prices here. So why do investors hate the traders
providing this unprecedented cost-reduction benefit?
"If you run the actual
numbers, trading has never been cheaper or easier for
the retail investor"
The answer, ironically, may lie in the very success of NMS
at reducing costs. HFT is the apotheosis of the SEC's
decades-long crusade to apply fairness to the market by
forcibly automating the old human trading processes. The
problem is, once you get onto the fairness treadmill, the
demands for more fairness will never cease. As it
asymptotically approaches the perfection of middleman-free
trading, the SEC will always hear gripes about the advantages
middlemen still have, which will never seem anything but unfair
In this atmosphere there are no cures for HFT that can make
a positive difference in the way it is perceived. Not assessing
cancellation fees, or requiring a minimum time on the book
before cancellation, or imposing market making obligations.
None of these will make any difference in the perceived
unfairness of HFT, and all would be uselessly disruptive as the
markets once again have to deal with major structural
Moreover, none of these reforms would address the real
problem this obsession with fairness has caused. Namely, that
the capital pipeline of new initial public offerings (IPOs) has
virtually dried up as the human middlemen who once made raising
capital their mission in life were wiped off the board by NMS.
There may be some way to raise capital on level playing fields,
but no one has found it yet, including the SEC.
So, again, what should we think about cartels of middlemen
creating monopolies to jack up prices? Back in 1792, when what
became the NYSE was taking shape as a price fixing cartel
called the Buttonwood Agreement, the main thing the public
wanted was not to prevent the conspiring, but to get a piece of
the action. The newly free and ambitious Americans just wanted
in on the speculative opportunities the Buttonwood brokers were
offering. After all, 1792 was only 16 years after the
Declaration of Independence established the right to be, among
other things, greedy middlemen or speculators.
'Speculation is the essential native genius of Wall Street,'
according to one historian and former SEC director of the
Office of Policy Research writing years later about the
Buttonwood period, (Walter Werner, Wall Street, 1991).
Why did we ever think creating the SEC to rein in speculation
was a good idea? Sure, there was the crash of 1929 and the
depression. But that might just prove the old adage: act in
haste, repent in leisure. In any case, why do we still think
today that letting the Commission destroy the structure of Wall
Street by imposing fairness will improve its functions?
The essence of Buttonwood was the unfair separation of
members from non-members, as the insiders pledged to maintain
that separation and to help each other exploit its advantages.
Back then, getting ahead was not a crime. The last thing we
should be worrying about now is leveling the playing field
between HFTs and regular investors. Focusing on this mission
impossible is only giving the SEC further licence to expand its
capital-destruction agenda. Better to just enjoy the
nearly-free trading provided by HFT, and worry about how to get
the SEC and other regulators out of the way of raising capital
before the economies of the west drop further into their
By Steve Wunsch, a New York-based writer and market
structure consultant. His latest book, Nature's God, is
available as a paperback or Kindle e-book.