Prewitt FTP 2M.

doc 12/12/2012 8:31 AM

NOTE

HIGH-FREQUENCY TRADING: SHOULD

REGULATORS DO MORE?
Matt Prewitt*

Cite as: Matt Prewitt, High-Frequency Trading: Should Regulators Do More?,

19 Mich. Telecomm. & Tech. L. Rev. 131 (2012), available at
http://www.mttlr.org/volnineteen/prewitt.pdf

High-Frequency Trading (“HFT”) is a diverse set of algorithmic trad-

ing strategies characterized by fast order execution. Its importance in
international markets has increased vastly in recent years.1 From a
regulatory perspective, HFT presents difficult and partially unresolved
questions. The difficulties stem partly from the fact that HFT encom-
passes a wide range of trading strategies, and partly from a dearth of
unambiguous empirical findings about HFT’s effects on markets. Yet
certain important conclusions are broadly accepted. HFT can increase
systemic risk by causing or exacerbating events like the “Flash Crash”
of May 6, 2010. HFT can also enable market manipulators to go unde-
tected. Finally, HFT’s supposed benefits to market quality (i.e., the
complex and interrelated topics of liquidity, volatility, and price dis-
covery) are questionable. Overall, the empirical research does not
demonstrate that HFT has substantial social benefits justifying its clear
risks. Regulatory measures including stronger monitoring, order can-
cellation taxes, and resting rules deserve more urgent attention.

Introduction ............................................................................................ 132

I. What is High-Frequency Trading? ......................................... 134
A. Strategies ................................................................................ 134
B. Methodologies ........................................................................ 137
II. The Effects of High-Frequency Trading
on the Market............................................................................ 139
A. Liquidity ................................................................................. 139
B. Volatility ................................................................................. 141
1. Anomalous Volatility ....................................................... 141
2. Normal Volatility ............................................................. 142

* Matt Prewitt is a third­year student at the University of Michigan Law School.

Many thanks to Professor Michael Barr, and to the editors of MTTLR, for their valuable assis-
tance with this Note.
1. Jeremy Grant, High­Frequency Trading: Up Against a Bandsaw, Fin. Times (Sept.
02, 2010), http://www.ft.com/cms/s/0/b2373a36-b6c2-11df-b3dd-00144feabdc0.html (esti-
mating that, in 2010, 56 percent of the equity trades in the United States and 38 percent in
Europe were executed by high-frequency traders).

131
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

132 Michigan Telecommunications and Technology Law Review [Vol. 19:131

C. Price Discovery and Market Efficiency .................................. 143

D. Cross-Market Propagation, Systemic Risk, and
Market Resiliency ................................................................... 146
E. Manipulation and Market Integrity ........................................ 147
III. Regulatory Responses .............................................................. 148
A. Three Complementary Perspectives for Regulators................ 149
1. High-Frequency Trading as a Systemic Risk................... 151
2. High-Frequency Trading as a Locus of
Illegality or Deceptive Practices ...................................... 155
3. High-Frequency Trading as a Detriment
to Day-to-Day Market Quality ........................................ 159
Conclusion ............................................................................................... 160

Introduction
High-Frequency Trading (“HFT”) has deservedly captured the attention
of both regulators and the public. For some, HFT exemplifies the proposi-
tion that a financial elite is earning fortunes with socially useless
techniques.2 For others, HFT simply reflects the logical progression of tech-
nology in which markets operate with increasing speed, precision, and
efficiency.3 The truth lies between these caricatures; however, given the im-
portance of HFT in modern markets, it is essential that regulators work
assiduously to grasp the issues and ensure that these dynamic technologies
do not cause unintended problems.
On May 6, 2010, HFT earned its place on the regulatory agenda when
the Dow Jones Industrial Average lost nearly one thousand points in just a
matter of minutes.4 After the smoke cleared, it became apparent that no ter-
rorist attack, sovereign default, mega­bankruptcy, or other fundamental
event had occurred. Indices and stocks quickly recovered and closed the day
down only about 3 percent.5 But in an investigation of that frightening dislo-
cation, the staffs of the Commodity Futures Trading Commission (“CFTC”)
and the Securities Exchange Commission (“SEC”) determined that the sud-
den volatility had coincided with HFT activity.6 A comprehensive report
released several months later detailed the chain of events on May 6 and con-

2. See generally Charles Duhigg, Stock Traders Find Speed Pays, in Milliseconds,
N.Y. Times (July 23, 2009), http://www.nytimes.com/2009/07/24/business/24trading.html.
3. High-Frequency Trading, The Economist, http://www.economist.com/debate/
overview/224 (last visited Oct. 22, 2012) (noting that 55 percent of The Economist’s online
voters believe that high-frequency trading contributes to the overall quality of markets).
4. Bruno Biais & Paul Woolley, High Frequency Trading 13 (Mar. 2011) (un-
published manuscript), available at http://idei.fr/doc/conf/pwri/biais_pwri_0311.pdf.
5. Staffs of the Commodity Futures Trading Comm’n & Sec. Exch. Comm’n,
Findings Regarding the Market Events of May 6, 2010, at 1–3 (2010), available at
http://www.sec.gov/news/studies/2010/marketevents-report.pdf [hereinafter CFTC-SEC Find-
ings].
6. Id.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 133

cluded that HFT played a key role in exacerbating the markets’ rapid
downward movements.7 By this time, it became clear to global regulators
that HFT deserved their attention.
Several market events during the summer of 2012 renewed concerns
about HFT. First, during Facebook’s initial public offering on May 18, a
high volume of rapid order cancellations overwhelmed NASDAQ’s comput-
er systems.8 While the incident has not yet been comprehensively studied,
HFT may have contributed to these technical problems, which interfered
with many traders’ orders.9 Then, on August 1, HFT market maker Knight
Capital caused rapid price movements in 150 NYSE stocks when one of its
algorithms malfunctioned.10 It appears that Knight’s algorithm uncontrolla-
bly bought high and sold low, losing $440 million, causing irrational swings
in affected stocks, and shaking the public’s confidence in the integrity of the
market.11
But how exactly does HFT work, and what issues does it raise? Have
regulators properly evaluated these issues and taken steps to protect
markets?12 Section I of this Note summarizes the key strategies and method-
ologies that constitute HFT. Section II traces the thorny academic questions
surrounding how HFT affects markets. Section III examines whether regula-
tors in Europe and the United States have properly assessed the problem and
taken the right regulatory steps. The Section further argues that regulators
are moving in a good direction but nevertheless ought to place broader re-
strictions on HFT. In its conclusion, this Note argues that such broader
restrictions—like cancellation taxes, transaction taxes, or resting rules—
would mitigate HFT’s proven downsides, but at the cost of HFT’s specula-
tive and unproven benefits.

7. Id.
8. Michael J. De La Merced, Nasdaq Concedes Facebook Missteps, N.Y. Times (May
20, 2012), http://dealbook.nytimes.com/2012/05/20/nasdaq-chief-says-glitches-werent-at-
fault-for-facebook-stock-plunge/.
9. See, e.g., Ivy Schmerken, Facebook: The Strangest IPO of All Time?, Wall St. &
Tech. (May 24, 2012), http://www.wallstreetandtech.com/exchanges/facebook-the-strangest-
ipo-of-all-time/240000996?pgno=1; see also How HFT Caused the Opening Delay, and Later
Benefited at the Retail Customer’s Expense, Nanex (May 18, 2012), http://www.nanex.net/
aqck/3099.html.
10. Knightmare on Wall Street, Nanex (Aug. 13, 2012), http://www.nanex.net/aqck2/
3522.html.
11. Id.
12. It is possible to conceive of HFT as a prudential issue in which regulators should
ensure that HFT firms do not incur systemically harmful losses. From this viewpoint, HFT is
merely a special case in the larger question of how to set capital and prudential requirements
for hedge funds and other proprietary traders. This Note will focus primarily on the following
inquiries: What are HFT’s externalities in the market? How does the high volume of HFT on
the market affect other participants? The answers to these questions reveal that HFT intimate-
ly affects general market quality, and therefore regulators should not view it as merely a risk
to the firms that use it.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

134 Michigan Telecommunications and Technology Law Review [Vol. 19:131

I. What is High-Frequency Trading?

HFT is an umbrella term referring to a diverse set of strategies whose
common denominator is that they are algorithmic and attempt to use low
latency (i.e., fast order execution) to gain an edge in the market. High-
Frequency Traders (“HFTs”) place many, if not most, of the trades on
today’s equity markets.13 HFTs have achieved extraordinarily low latency,
meaning that very little time elapses between when they send orders and
when the orders are executed.14 Top HFTs achieve latency of only a few
thousandths of a second, such that millionths of a second are increasingly
becoming the pertinent measure of latency.15 Because additional increments
of latency can be the difference between executing profitable trades and ced-
ing opportunities to faster HFTs, the race to “zero latency” will likely
continue unabated.
A range of institutions use HFT. Some hedge funds, like Citadel and
Renaissance, make HFT a prominent part of their investing strategy.16 Other
firms, like Getco, focus exclusively on HFT strategies.17 Banks engaging in
proprietary trading have also used HFT, but the Volcker rule will diminish
that activity by reducing banks’ overall levels of proprietary trading.18 Alt-
hough traditional “buy­side” investors19 do not generally use HFT, many
institutional investors use order-execution services offered by HFT firms in
order to optimize the price received or to escape detection by counterparties
who want to avoid trading with them.20

A. Strategies
HFT strategies are diverse, proprietary, and complex, so it is not possi-
ble to describe them except at a somewhat unfortunate level of abstraction.
Each strategy is susceptible to innumerable nuances, some of which I will
describe in the following Sections. Furthermore, HFTs operate in diverse

13. See Grant, supra note 1.

14. Andrew G. Haldane, Exec. Dir. of Fin. Stability, Bank of Eng., Speech to the Int’l
Econ. Ass’n Sixteenth World Congress: The Race to Zero 5 (July 8, 2011), available at
http://www.bankofengland.co.uk/publications/Documents/speeches/2011/speech509.pdf.
15. Id.
16. Biais & Woolley, supra note 4, at 3.
17. Id.
18. See Dodd-Frank Act § 619, 12 U.S.C. § 1851 (2011). The Volcker Rule limits tra-
ditional banks’ ability to trade with their own money.
19. “Buy-side” generally refers to institutions like mutual funds or pension funds which
purchase large blocks of securities, typically for wealth management.
20. For example, Getco offers a service called “GETAlpha,” which it describes on its
website as follows: “GETAlpha is a customizable suite of trading tools built to capture ad-
vantages across rapidly changing markets, giving institutional investors a range of trading
strategies expressly designed to navigate the complexities of today’s multi-venue market-
place . . . . GETAlpha offers the investment community maximum liquidity with minimum
detection.” GETAlpha, GETCo, http://www.getcollc.com/GES/index.php/our_offerings/
GETAlpha/ (last visited Nov. 23, 2012).
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 135

markets, from foreign exchanges (“FX”) to derivatives and equities. None-

theless, the principal strategies fall under this rubric: market making,
momentum or event trading, liquidity detection, and arbitrage.
In market making, HFTs act like a faster version of traditional market
makers who buy and sell securities in order to profit from the difference, or
spread, between bid and ask prices.21 In some cases, exchanges subsidize
this type of trading because it makes trading easier for all participants.
These subsidies come in the form of “rebates,” or reduced transaction fees,
which bolster the profitability of each liquidity-providing trade.22 Speed
provides an advantage in capturing the spread, because fast trades are less
likely to be affected by price movements. A further difference between
HFTs and traditional market makers is that traditional market makers have
agreements with exchanges to continue providing liquidity even when they
would rather not (for example, when the market is rapidly falling). In con-
trast, liquidity­providing HFTs sometimes do not participate in official
market­making programs promoted by the exchanges. Such traders have the
option of leaving the market, and thus ceasing their liquidity provision, at
any time.23
In momentum or event trading, HFTs behave analogously to day trad-
ers. HFT algorithms use a variety of techniques to predict short­term price
movements and place marketable orders in the direction of the movement.
For example, momentum trading involves identifying price movements that
are likely to persist in the short term, then trading directionally while the
movement continues and ceasing when it stops.24 HFTs may also make pre-
dictions involving statistical phenomena like mean reversion, which is the
theory that prices tend to gravitate toward historically average levels.25
Mean­reversion trading therefore involves betting that large deviations from
historical average prices will not persist. Finally, HFTs may engage in event
trading by betting on market responses to new information like economic
data releases from the government or the Federal Reserve.26
In liquidity detection, HFT algorithms attempt to identify and profit
from the actions of other large traders. For example, by aggregating multiple

21. Peter Gomber et al., High­Frequency Trading 25 (Mar. 2011) (unpublished manu-
script), available at http://ssrn.com/abstract=1858626.
22. Id. Markets compete with each other in providing rebates in order to attract liquidi-
ty providers.
23. See Brian Weller, Liquidity and High Frequency Trading 7 (Nov. 10, 2012) (un-
published manuscript), available at http://home.uchicago.edu/~bweller/files/Liquidity_
and_High_Frequency_Trading.pdf (discussing the categorization of fast traders who are not
market makers as HFTs). For further discussion of HFT market making, see Albert J. Men-
kveld, High Frequency Trading and the New-Market Makers 27 (Feb. 6, 2012) (unpublished
manuscript), available at http://ssrn.com/abstract=1722924.
24. Gomber et al., supra note 21, at 30.
25. Bank for Int’l Settlements, High-Frequency Trading in the Foreign Ex-
change Market 5 (2011), available at http://www.bis.org/publ/mktc05.pdf.
26. Id.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

136 Michigan Telecommunications and Technology Law Review [Vol. 19:131

data points from different exchanges and looking for characteristic patterns
in variables like order depth, HFTs may determine the existence of a large
hidden limit order or a large trader attempting to enter or exit a position.27
Next, HFTs attempt to profit from the price movement created by other
traders; for example, HFTs may buy just prior to the execution of other trad-
ers’ large marketable orders.28 A variant of this strategy involves attempting
to detect and predict the behavioral patterns of other algorithmic traders and
exploiting their impact on the market.29
Much HFT activity belongs in the broad category of arbitrage. Arbitrage
involves identifying two or more securities that structurally tend to move in
unison.30 When they fall out of alignment, arbitrageurs buy the cheaper one
and sell the more expensive one until the difference is eliminated.31 Due to
its speed, HFT can engage in arbitrage involving extremely short time
frames and, consequently, extremely small price differences.32 This makes it
possible to profit from miniscule misalignments, for example, between iden-
tical assets on different exchanges. 33 HFT arbitrageurs can trade on
misalignments between different markets, between derivatives and their un-
derlying assets, between exchange­traded funds (“ETFs”) and their
constituent securities, or simply between statistically correlated assets on the
same market.34 “Latency arbitrage” involves trading in the sub­second time
windows between when market prices move and when market makers up-
date their quoted prices. During these time windows, HFTs have more
information than slower traders, which allows them to profit at slower trad-
ers’ expense.35
Unfortunately, these descriptions of HFT strategies come nowhere close
to exhausting the topic; the possibilities and nuances surrounding each strat-
egy are endless. It is important to note, however, that most of these
strategies (in their high­frequency incarnations) involve not only fast order-
ing but also fast order cancellation.36 HFT market makers, for example,
constantly cancel orders to optimize their quotes and avoid entering into
trades not informed by up­to­the­millisecond information. 37 Arbitrageurs

27. Gomber et al., supra note 21, at 28–29.

28. Id.
29. Id.
30. Biais & Woolley, supra note 4, at 6.
31. Id.
32. Id.
33. Gomber et al., supra note 21, at 27.
34. Id. at 27–28.
35. See Bank for Int’l. Settlements, supra note 25, at 5.
36. Indeed, some researchers suggest using cancellation levels as a proxy for measuring
HFT activity. See Bruno Biais, Thierry Foucault, & Sophie Moinas, Equilibrium High Fre-
quency Trading 28 (Sept. 2011) (unpublished manuscript), available at http://ssrn.com/
abstract=1834344.
37. Int’l Org. of Sec. Comm’ns, Regulatory Issues Raised by the Impact of
Technological Changes on Market Integrity and Efficiency 23 (2011).
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 137

and directional traders may also use rapid, frequent order cancellation for a
range of purposes. For example, liquidity detection often involves sending
out and immediately cancelling orders in order to gain information about
invisible liquidity lurking off the public ticker (this is sometimes called
“pinging”).38 If the ping results in a trade before it is cancelled, the HFT can
use that information to infer the existence of a liquidity provider.

B. Methodologies
Familiarity with basic HFT methodologies helps one understand the
toolbox available to HFTs and the current HFT environment more generally.
In the immediate wake of the Flash Crash, U.S. regulators quickly took steps
that served to limit some of the market-access methodologies available to
HFTs. Regulations ensuring fairness in HFTs’ methods of obtaining market
access are, from a regulatory perspective, low-hanging fruit. Market partici-
pants have largely embraced these regulations,39 and while the changes cannot
eliminate fundamental concerns about HFT, they can meaningfully improve
fairness and risk.
Prior to the Flash Crash, many HFT firms gained special access to ex-
changes using a technique called “naked access.” With naked access,
brokers allowed HFTs to essentially piggyback on the brokers’ direct access
to markets.40 This permitted HFTs to reduce their trade latency while also
avoiding the risk checks and capital requirements to which they would be
subject if they were direct members of the market.41 In November 2010, the
SEC issued a new rule directed at brokers, Rule 15c3-5, that made this risk-
exacerbating practice impossible.42 The proposed revisions of the Markets in
Financial Instruments Directive (“MiFID”) would have the same effect in
Europe.43
A second important tool for understanding the HFT environment is
co­location. In the late 1990s through the early 2000s, many electronic ex-
changes began allowing firms to locate their servers at the same facility as
the exchanges’ servers.44 This allows HFTs to achieve lower latency. It thus

38. Jaksa Cvitanic & Andrei Kirilenko, High Frequency Traders and Asset Prices 2
(Mar. 11, 2010) (unpublished manuscript), available at http://ssrn.com/abstract=1569075.
39. Gomber et al., supra note 21, at 41.
40. Id.
41. Id.
42. 17 C.F.R. § 240.15c3-5 (2012).
43. Proposal for a Directive of the European Parliament and of the Council on Mar-
kets in Financial Instruments Repealing Directive 2004/39/EC of the European Parliament
and of the Council, at 25–26, COM (2011) 656 final (Oct. 20, 2011), available at http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2011:0656:FIN:EN:PDF [hereinafter
Proposed Directive] (“It is desirable to ensure that all high frequency trading firms be author-
ised when they are a direct member of a trading venue. This should ensure they are subject to
organisational requirements under the Directive and are properly supervised.”).
44. See, e.g., Graham Bowley, The New Speed of Money, N.Y. Times (Jan. 1, 2011),
http://www.nytimes.com/2011/01/02/business/02speed.html?pagewanted=all.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

138 Michigan Telecommunications and Technology Law Review [Vol. 19:131

confers advantages over other fast traders who are not co­located. Out of
fairness concerns, the CFTC proposed a rule in June 2010 requiring uniform
fees and access to co­location facilities.45 The proposed new version of Mi-
FID would likewise require equitable co­location practices.46
Perhaps the most controversial form of privileged market access for
HFTs is the “flash order.” Flash orders are particularly pertinent in the U.S.
due to SEC Regulation NMS, which requires orders to be routed to the ex-
change offering the best price.47 A flash order is a marketable order that,
immediately prior to being rerouted in accordance with Regulation NMS, is
flashed for milliseconds on the exchange where it is initially placed.48 Be-
cause flash orders persist for only milliseconds, regular traders cannot place
trades against them before they are withdrawn. HFTs, on the other hand,
sometimes act quickly enough to execute against flash orders. Thus, observ-
ers have voiced concern about creating a “two­tier” market in which HFTs
could trade amongst themselves, increasing their informational advantage
over slow traders.49 The SEC proposed eliminating the rule exception per-
mitting flash orders in 2009, but has not finalized that change.50 While some
smaller exchanges have held out, most major exchanges have voluntarily
stopped the practice. For example, Direct Edge stopped offering flash orders
for stock trading in 2011, but apparently continued to allow flash orders for
options trading.51 Some smaller exchanges, like the Chicago Board Options
Exchange (“CBOE”), have continued to allow flash orders on certain kinds
of trades.52

45. Co-Location/Proximity Hosting Services, 75 Fed. Reg. 33198 (proposed June 11,
2010); Gomber et al., supra note 21, at 43.
46. Proposed Directive, supra note 43, at 117 (“Member States shall require a regulat-
ed market to ensure that its rules on co-location services and fee structures are transparent, fair
and non-discriminatory.”).
47. 17 C.F.R. § 242.602(a)(1)(i) (2012) (“Each national securities exchange shall at all
times such exchange is open for trading, collect, process, and make available to vendors the
best bid, the best offer, and aggregate quotation sizes for each subject security listed or admit-
ted to unlisted trading privileges which is communicated on any national securities exchange
by any responsible broker or dealer, but shall not include: (A) Any bid or offer executed im-
mediately after communication and any bid or offer communicated by a responsible broker or
dealer other than an exchange market maker which is cancelled or withdrawn if not executed
immediately after communication . . . .”).
48. Gomber et al., supra note 21, at 42.
49. Id.
50. Elimination of Flash Order Exception from Rule 602 of Regulation NMS, Ex-
change Act Release No. 34-60684, 74 Fed. Reg. 48632 (Sept. 23, 2009), available at
http://www.sec.gov/rules/proposed/2009/34-60684.pdf. For the Rule 602(a)(1)(i)(A) excep-
tion permitting flash orders, see 17 C.F.R. § 242.602 (2012).
51. Jacob Bunge, Direct Edge to Stop “Flashing” Orders on Monday, Wall St. J.
(Feb. 25, 2011), http://online.wsj.com/article/SB10001424052748703409304576166930877
474292.html.
52. See NBBO Step-Up, Chi. Bd. of Exch., https://www.cboe.org/hybrid/nbbosu.aspx
(last visited Oct. 25, 2012) (describing the CBOE’s National Best Bid and Offer “Step Up”
system, which allows orders to be flashed for 150 milliseconds).
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 139

II. The Effects of High-Frequency

Trading on the Market
While proponents argue that HFT is a socially beneficial liquidity pro-
vider,53 many regulators and scholars worry that HFT harms various aspects
of market functioning54 and that HFT’s liquidity benefits have been overstat-
ed.55 This Section will summarize and evaluate the small but growing body
of academic research pertaining to HFT’s effect on markets. It will further
discuss five distinct yet interrelated market qualities that HFT has been
thought to influence: liquidity, volatility, price discovery, market resiliency,
and market integrity.

A. Liquidity
A substantial body of literature suggests that HFT supplies liquidity to
markets.56 Liquidity refers generally to the ease of transacting; in this con-
text, it is useful to think of liquidity as the ability to find ready buyers and
sellers at or near the prevailing market price of a given security.57 Uncontro-
versially, HFT can provide liquidity, such as by intermediating large orders:
algorithms break large orders into pieces and rapidly find smaller buyers or
sellers willing to transact at a price close to the prevailing market price.58 By
contrast, human intermediaries are slower and might need to find larger
counterparties, potentially exposing themselves to delays and larger spreads.

53. Terrence Hendershott et al., Does Algorithmic Trading Improve Liquidity?, 66 J.

Fin. 30, 30–31 (2011), available at http://faculty.haas.berkeley.edu/hender/Algo.pdf;
Menkveld, supra note 23, at 27–28; Ray Riordan & Andreas Storkenmaier, Latency, Liquidity
and Price Discovery, 20–21, 15 J. Fin. Mkts. 416 (2012).
54. See generally X. Frank Zhang, High Frequency Trading, Stock Volatility, and Price
Discovery 33–35 (2010) (unpublished manuscript), available at http://papers.ssrn.com/sol3/
papers.cfm?abstract_id=1691679; Public Consultation: Review of the Markets in Financial
Instruments Directive (MiFID), at 14–17, COM (2010), available at http://ec.europa.eu/
internal_market/consultations/docs/2010/mifid/consultation_paper_en.pdf; see also CFTC-
SEC Findings, supra note 5.
55. Andrei A. Kirilenko et al., The Flash Crash: The Impact of High Frequency Trad-
ing on an Electronic Market 37–38 (May 26, 2011) (unpublished manuscript), available at
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1686004.
56. See generally Hendershott et al., supra note 53; Riordan & Storkenmaier, supra
note 53.
57. Usually, liquidity is thought of as being “supplied” by limit orders, which execute
only after the market price moves to a predetermined point, and “taken” by market orders that
execute at whatever price is currently on offer, thereby diminishing depth on the other side of
the book and potentially moving the price. HFT uses both kinds of orders routinely. See Joel
Hasbrouck & Gideon Saar, Low-Latency Trading 12 (Feb. 2011) (unpublished manuscript),
available at http://www.bus.umich.edu/academics/departments/finance/Sem%20Papers/W%
202011%20Hasbrouck.pdf. However, algorithmic trading somewhat confounds this traditional
way of thinking about liquidity, because some algorithms use limit orders in ways that effec-
tively diminish liquidity. Id.
58. Hendershott et al., supra note 53, at 2.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

140 Michigan Telecommunications and Technology Law Review [Vol. 19:131

HFTs have thus been analogized to a more efficient version of the

pre­digital market maker or floor specialist.59
However, HFT does not always play this liquidity­supplying function.
HFT firms may engage in rapid liquidity­taking trades or abruptly stop sup-
plying liquidity.60 Additionally, lower spreads caused by HFT’s liquidity
provision are partially illusory because HFT also tends to provide low order
depth. 61 In other words, HFT market makers sometimes quote narrow
spreads without being willing to buy or sell substantial quantities at those
prices. Therefore, the low spread does not last if anyone tries to transact a
significant quantity. Compounding these issues, HFT displaces other kinds
of liquidity suppliers, like exchange-certified specialists—legacy market
makers who might have been more likely to stay in the market during turbu-
lence, thus preserving market liquidity and stability.62
HFT may also cause liquidity to dry up for more complex reasons.
During the Flash Crash, the pressure of a single large sell order in the E-
Mini—a major S&P 500 futures index—caused HFTs to acquire large
blocks of E-Mini shares, which they then rapidly unloaded to escape their
net­long position.63 However, the only parties willing to rapidly purchase
those shares were other HFT firms who had posted “stub quotes,” or unreal-
istically cheap limit orders to buy.64 These firms in turn resold to avoid
holding a net-long position. This meant that many orders were executed at
extremely cheap prices, so that the index fell rapidly, frightening traditional
liquidity suppliers and fundamental buyers out of the market.65 Under these
conditions, HFT’s creation of high trading volume did not correspond to a
provision of real, high­quality liquidity.66
Thus, an overarching concern emerges from the literature: HFT might
add liquidity to markets in good times, while having a negative effect when
market conditions are adverse or volatile. Not reassuringly, one of the most
important studies showing algorithmic trading’s positive effect on liquidity
was conducted during times of low volatility and rising prices.67 Nonethe-
less, the literature broadly indicates that HFT does not negatively affect

59. Menkveld, supra note 23, at 6.

60. Hasbrouck & Saar, supra note 57, at 36.
61. Hendershott et al., supra note 53, at 22.
62. Hasbrouck & Saar, supra note 57, at 31 (“In the face of transient supply and de-
mand, NYSE specialists were obligated to stabilize prices and maintain continuous presence
in the market. They were subject to restrictions on reaching across the market to take liquidity
(i.e., making destabilizing trades). Low­latency traders have no such obligations.”).
63. CFTC-SEC Findings, supra note 5, at 3.
64. Id. at 35–36.
65. Id. at 3–4.
66. Id. at 3.
67. Hendershott et al., supra note 53, at 31. This study shows that algorithmic trading
(a broader category that includes HFT) had the effect of narrowing spreads by examining the
periods before and after NYSE’s introduction of Autoquote—a tool that facilitated the entry
of many algorithmic traders. This occurred during the stable bull market of 2003.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 141

liquidity conditions on a day­to­day basis when trading conditions are nor-

mal. Because concerns about HFT’s effect on liquidity focus on the
possibility that HFT takes liquidity during periods of unusual or volatile
trading patterns,68 this issue is interrelated with worries about HFT’s effect
on volatility.

B. Volatility
Probably the most prominent concern about HFT is that it may exacer-
bate volatility, a concern intensified by the Flash Crash. Yet when
researchers discuss HFT’s effect on volatility, they sometimes have very
different conceptions of the term in mind.69 Much research has focused on
the kind of severe, sudden volatility exemplified by the Flash Crash (and
more recently, the price fluctuations caused by Knight Capital’s rogue algo-
rithm). I will call this “anomalous volatility.” Research on the Flash Crash
has established beyond serious dispute that HFT has the potential to create
anomalous volatility, as this Section will discuss. However, questions about
HFT’s effect on volatility under normal circumstances remain contested and
warrant further research attention. I will call this kind of volatility “normal
volatility.” This Section will examine HFT’s relationship with anomalous
volatility, followed by an examination of HFT’s effect on normal volatility.

1. Anomalous Volatility
Academic research surrounding the Flash Crash,70 in concert with a
joint report by the staffs of the SEC and CFTC,71 depicts a concrete situation
in which HFT exacerbated volatility initiated by an unusually large sell or-
der in the E-Mini. The chain of events on that day is complicated. But, most
notably, the SEC staff concluded that when HFT firms held too many shares
of the E-Mini at a moment of sub-normal fundamental demand, they played
“hot potato” by repeatedly selling to each other at very low prices, causing
an abnormal decline in the E-Mini index.72 HFT algorithms were transacting
on the buy side, despite having an unwanted long position, because they
were trying to take advantage of the liquidity that they expected other op-
portunistic buyers to supply in light of the sharply lower price.73 Cross­index
arbitrageurs and other liquidity suppliers did indeed buy the E-Mini while it
was low, but not in quantities sufficient to stop the downward spiral.74 Sim-
ultaneously, arbitraging HFTs sold the S&P 500 stocks to which the E-Mini
was linked, reinforcing the illusion of a fundamental market event that

68. Kirilenko et al., supra note 55, at 22.

69. Compare id., with Zhang, supra note 54.
70. See, e.g., Kirilenko et al., supra note 55, at 25–26.
71. CFTC-SEC Findings, supra note 5.
72. Id. at 3–4.
73. Id. at 15.
74. Id. at 5.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

142 Michigan Telecommunications and Technology Law Review [Vol. 19:131

scared fundamental buyers out of the market.75 The Flash Crash has become
the paradigmatic instance of HFTs causing or exacerbating anomalous vola-
tility.
Unusual market conditions triggered the Flash Crash: a sell order of rare
magnitude against a backdrop of deep anxiety about European defaults.76
Thus, the Flash Crash does not necessarily indicate that HFT has a general,
everyday volatility-exacerbating effect. Of course, it is cold comfort if HFT
only exacerbates volatility in situations of extreme stress, and the distinction
has important regulatory implications. If HFT exacerbates volatility only in
certain anomalous situations,77 it might make sense for regulators to focus
on preventing sudden crashes. But if HFT increases volatility more general-
ly, broader regulatory action curtailing HFT under everyday circumstances
might be justified.

2. Normal Volatility
Research conflicts on whether HFT increases volatility under normal
market conditions. At least one study has suggested that stocks traded heavi-
ly by HFTs are causally linked with greater volatility.78 Consistent with this
finding, officials from the Bank of England have pointed toward higher
cross­stock correlation as a proxy for HFT activity and have shown a positive
relationship between cross­stock correlation and volatility. 79 Nevertheless,
credible empirical research muddies this picture by showing that HFT de-
creases volatility in the short term. 80 Indeed, it makes sense that HFT’s

75. Id. at 16–18.

76. The crash was initiated by a kind of perfect storm: one of the largest sell orders of
the year, placed against a background of “thinning liquidity.” Id. at 2–3. Under normal cir-
cumstances, arbitrageurs presumably absorb anomalous movements caused by HFT.
77. David Easley et al., The Microstructure of the “Flash Crash”: Flow Toxicity, Li-
quidity Crashes, and the Probability of Informed Trading, J. Portfolio Mgmt., Winter 2011,
at 118. The authors suggest the creation of a new futures contract linked to a metric of the
probability of informed trading activity. They claim they have designed such a metric, which
would have registered unusual activity just prior to the Flash Crash. They argue such a con-
tract might help avoid future flash crashes for two reasons. First, HFTs could use it to hedge
against insufficient liquidity from informed traders (also known as adverse selection), as oc-
curred in the E-Mini during the Flash Crash. Therefore, at these dangerous moments, HFTs
might be willing to remain liquidity-providing market makers, instead of turning into liquidity
takers. Second, such a contract might give regulators or exchanges warning so that they could
shut down or slow trading when informed traders are providing dangerously little liquidity.
78. Zhang, supra note 54, at 34–35. Zhang establishes a correlation between HFT ac-
tivity and volatility, and supports a causal relationship between HFT and volatility by
examining the 2003 NYSE Autoquote introduction as a natural experiment.
79. Haldane, supra note 14, at 14.
80. Jonathan Brogaard et al., High Frequency Trading and Volatility 30 (July 30, 2012)
(unpublished manuscript), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_
id=1928510. Brogaard uses the temporary 2008 Short Sale Ban (with which the SEC banned
short sales in the wake of the 2008 crash) as an exogenous shock, discerns the amount of HFT
that was de facto banned by that measure, and then measures the resultant changes in volatili-
ty. He finds that HFT decreases intraday volatility. His study also indicates, however, that HFT
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 143

liquidity­providing function should dampen short­term volatility by making

it possible to buy and sell without significantly altering prices. However, the
research does not establish that HFT dampens volatility with any consisten-
cy, or that this short­term volatility dampening helps volatility in the longer
term.
Until the question is more clearly resolved, regulators will face hurdles
arguing that HFT creates general volatility costs (as opposed to anomalous
Flash Crash–like events). Research on the severe Eurodebt-related volatility
of summer and fall 2011, which some believe to have been exacerbated by
HFT,81 may eventually provide insight. But in the absence of more conclu-
sive research on HFT’s effect on everyday volatility, regulators are likely to
remain focused on exceptional volatility events analogous to the Flash Crash
or the Knight Capital rogue algorithm. There is little controversy over the
thesis that HFT can take liquidity and exacerbate volatility in these kinds of
anomalous circumstances.82

C. Price Discovery and Market Efficiency

Another important concern about HFT is that it damages price discov-
ery processes.83 In plain language, this is the worry that, either by introducing
unreliable information into prices or by making conditions difficult for traders
with sound information, HFT diminishes markets’ ability to incorporate
information into share values.84 Price discovery is important because one of
the most critical functions of public markets is communicating reliable eco-
nomic information—even to non-market participants—through price levels.

becomes a liquidity taker when macro (i.e., not stock­specific) news induces market volatility,
perhaps because macro news is hard to hedge against. This aspect of Brogaard’s finding may
therefore lend support to the hypothesis that HFT contributed to the extreme macro news–
driven volatility of summer and fall 2011. Hasbrouck & Saar, supra note 57, at 36, find that
HFT decreases short-term volatility. But see Brogaard et al., supra, at 5 (noting that the
Hasbrouck & Saar study only looks at order book activity, an indirect proxy for HFT).
81. A sharp rise in the Chicago Board Options Exchange Market Volatility Index
(“VIX”) was fundamentally driven by uncertainty about the future of European economies,
but may have been amplified by HFT. See Off the Charts: Excess Stock Market Volatility,
N.Y. Times (Nov. 4, 2011), http://www.nytimes.com/interactive/2011/11/04/business/economy/
Off-the-charts-excess-stock-market-volatility.html.
82. See Brogaard et al., supra note 80, at 30. Brogaard’s study is perhaps the most­cited
empirical work suggesting that HFT decreases intraday volatility. However, his study is con-
sistent with the theory that HFT exacerbates volatility under turbulent circumstances. He finds
that although HFT activity overall reduces net intraday volatility, HFT exacerbates volatility
following macroeconomic news­induced movements.
83. Int’l Org. of Sec. Comm’ns, supra note 37, at 27.
84. It is important to note the interconnections between questions about HFT’s role in
price discovery, liquidity, and volatility. As explained supra note 57, a traditional supplier of
liquidity places limit orders. Limit orders allow transactions to take place without a change in
the price. Liquidity suppliers tend to dampen a price’s response to new information. Marketa-
ble orders, on the other hand, are generally understood as facilitating price discovery.
Accordingly, Hendershott et al., supra note 67, at 23 conclude that HFTs aid price discovery
through their marketable orders.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

144 Michigan Telecommunications and Technology Law Review [Vol. 19:131

Rightly or wrongly, the issue of HFT’s potential harm to price discovery has
generated less public concern than the issue of volatility. Some regulators
accept the argument that HFT helps price discovery.85 Nonetheless, many
academics consider HFT’s effect on price discovery to be a key concern.86
HFT might either harm or help price discovery processes for several in-
tuitive reasons, and it is helpful to keep these in mind. First, HFTs hold their
securities for short periods of time to avoid exposure to fundamental­driven
price movements87 and therefore have little interest in the fundamental value
of the securities they trade. This suggests that HFTs do not contribute new
information to security prices, unlike long-term investors who carefully ana-
lyze the underlying value of assets. On the other hand, HFTs outperform
traders not using high-frequency strategies in some price discovery contexts.
Many HFTs are, in one form or another, arbitrageurs.88 Thus, when asset A’s
value is affected by changes in the price of asset B, HFTs incorporate this
information more quickly and accurately than slower human traders.89
Different HFT strategies, just like various traditional trading strategies,
have differing effects on price discovery. This makes it problematic to study
the price discovery implications of HFT as an umbrella category.90 Unfortu-
nately, much of the leading empirical research on price discovery research
does just that:91 its findings pertain to HFT’s aggregate contribution to price
discovery. This creates interpretive difficulties. The problem is that an over-
all positive effect of HFT on price discovery might hide the fact that some
subset of HFT harms price discovery.92 For example, even if cross­market
price arbitrage helps price discovery, some statistical arbitrage strategies
might contribute unreliable information to prices.93 Further, market-making

85. Eur. Sec. & Mkt. Auth., Consultation Paper: Guidelines on Systems and Controls
in a Highly Automated Trading Environment for Trading Platforms, Investment Firms and
Competent Authorities 51, ESMA/2011/224 (July 20, 2011), available at http://www.
esma.europa.eu/system/files/esma_2012_122_en.pdf.
86. See, e.g., Zhang, supra note 54, at 1–2; Biais & Woolley, supra note 4, at 14–15.
87. Kirilenko et al., supra note 55, at 17.
88. See supra notes 30-35 and accompanying text.
89. Biais & Woolley, supra note 4, at 6. Arbitrage strategies also target differing price
movements in the same asset on different exchanges, triangular arbitrage as in FX trading, and
more complex strategies.
90. Zhang, supra note 54, at 10 (“A tick by tick study using open market data is likely
to be influenced by HFT’s market making activities, which tend to be more beneficial to the
capital market than aggressive HFT strategies.”).
91. See id. at 3–4; see also Jonathan Brogaard, High Frequency Trading and Its Impact
on Market Quality 6 (July 16, 2010) (unpublished manuscript), available at
http://www.futuresindustry.org/ptg/downloads/HFT_Trading.pdf. One partial exception is
Jonathan Brogaard et al., High Frequency Trading and Price Discovery 26–28 (July 30, 2012)
(unpublished manuscript), available at http://ssrn.com/abstract=1928510, which separately
analyzes the price discovery implications of HFT’s limit orders and marketable orders.
92. See Zhang, supra note 54, at 3.
93. While cross­market arbitrageurs trade on information about the same asset in other
fora, statistical arbitrage strategies trade on more speculative relationships between different
assets. See supra notes 30–36 and accompanying text.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 145

HFT strategies have wholly different price discovery implications.94 Re-

searchers who treat HFT as a monolithic whole do so due to the
impracticability of distinguishing between strategies in the datasets, but reg-
ulators have the option of regulating different HFT strategies differently.95
Regulators should pay attention to this interpretive problem. Nonetheless,
regulators can gain considerable insight from existing studies.
Empirical findings conflict regarding HFT’s aggregate contribution to
price discovery. However, the view advanced in Jonathan Brogaard’s research
has been broadly endorsed.96 Brogaard suggests that price movements initiat-
ed by HFT have a more lasting effect than price movements initiated by
non-HFT, indicating that HFT helps price discovery.97 Brogaard, Hendershott,
and Riordan lend support to this view, finding that HFT’s marketable orders
tend to move in the direction of non­transitory price changes.98 On the other
hand, X. Frank Zhang finds that—consistent with traditional (i.e., pre-HFT)
theories about short-run trading—HFT causes prices to overreact to news
about a company’s fundamentals.99 Yet Zhang’s findings do not necessarily
contradict those of Brogaard, or of Hendershott and Riordan, because Zhang
purports to examine price efficiency over longer time frames than his col-
leagues. 100 Zhang’s study examines price efficiency over the course of
months, whereas his colleagues look at price efficiency within the day.101
Acknowledging this distinction, Hendershott and Riordan raise a serious
concern that Zhang’s methodology may be overinclusive, capturing the ef-
fects of non-HFT short­term trading.102
The scholarly debate over HFT’s effect on price discovery therefore
continues. But the research proclaiming that HFT helps price discovery
should fail to ease regulators’ minds. What do these findings really mean?

94. Gomber et al., supra note 21, at 59; Zhang, supra note 54, at 10.
95. In September 2011, the SEC and FINRA requested computer code used in trading
from several HFTs in order to examine it in detail. Although most observers believe this re-
quest was aimed at finding illegal or market-manipulating algorithms, it was also a potential
first step towards differentiating between different HFT strategies in regulation. See Sarah N.
Lynch & Jonathan Spicer, U.S. Regulators Seek High-Frequency Trading Secrets, Reuters
(Sept. 2, 2011, 9:34AM), http://in.reuters.com/article/2011/09/02/idINIndia-59107920110902.
96. See, e.g., Eur. Sec. & Mkt. Auth., supra note 85, at 63; Bank for Int’l Settle-
ments, supra note 25, at 13.
97. See Brogaard, supra note 91, at 46, 53.
98. Brogaard et al., supra note 91, at 2. They also find that HFTs’ limit orders, which
have less of an effect on price, tend to lose money and to execute against informed counter-
parties. Id. Costs are recouped in these trades through the bid-ask spread and liquidity rebates.
Id.
99. Zhang, supra note 54, at 26.
100. Zhang examines accumulated effects on price discovery over the course of quarters.
Id. at 9–10. Brogaard, Hendershott, and Riordan, on the other hand, examine whether HFT pri-
marily participates in short-term price movement versus utterly ephemeral “noise.” Brogaard et
al., supra note 91, at 2–3.
101. See Zhang, supra note 54, at 26.
102. Brogaard et al., supra note 91, at 4 n.5.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

146 Michigan Telecommunications and Technology Law Review [Vol. 19:131

Brogaard, Hendershott, and Riordan’s findings show that “HFT predicts

price movements for only tens of seconds.”103 If this short­term price dis-
covery is predicated on information that slower traders would otherwise
have soon acted upon, HFT’s price discovery function creates little or no
social benefit.104 On the contrary, it may discourage beneficial market partic-
ipation by trading ahead of market participants who have valid
information.105 Brogaard’s finding that HFTs successfully strive to provide
liquidity disproportionately to uninformed traders bolsters this possibility.106
The picture that emerges from the literature is one in which HFTs trade in
the same direction as price movements, but in doing so make trading cheap-
er for uninformed investors, and more expensive for investors who know
something about where the price ought to move. Regulators also cannot for-
get that this nominal contribution to short­term price discovery likely cloaks
subsets of HFT activity that are harmful to price discovery—or even out-
right manipulative, a subject further explored in Section II.E.

D. Cross-Market Propagation, Systemic Risk,

and Market Resiliency
Although the Flash Crash began with a plunge in the E-Mini, it did not
remain confined to that derivative alone. Rather, the plunge was rapidly mir-
rored in every major index and in individual company stocks. 107 This
happened even though no fundamental economic event triggered the plunge.
SEC and CFTC investigators concluded that the E-Mini plunge spread
rapidly to other stocks and indices through the activity of HFTs that auto-
matically arbitrage misalignments between related indices, and between
indices and the basket of stocks to which they correspond.108 This phenome-
non, sometimes called cross­market propagation, 109 raises concerns that
HFT may increase systemic risk by making markets less resilient to serious
price dislocations.
HFT may cause price dislocations to propagate through markets via var-
ious channels. As in the Flash Crash, HFT may send a shock spreading from
a derivative to its underlying assets. Similarly, HFT may rapidly propagate
price dislocations between similar stocks, between stocks and ETFs, and
between different trading platforms and exchanges.110
The flipside of this potential problem is that HFT helps investors deal
with fragmented markets by arbitraging prices between exchanges and

103. Id. at 19.

104. Brogaard, Hendershott, and Riordan note that they have no evidence as to whether
HFT incorporates information that slower humans would have incorporated anyway. Id.
105. Id.
106. Brogaard, supra note 80, at 11–16.
107. CFTC-SEC Findings, supra note 5, at 16–18.
108. Id.
109. Id. at 16.
110. Haldane, supra note 14, at 14.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 147

bringing liquidity supplies into alignment.111 Consistency across exchanges

may be particularly important in Europe in light of the market fragmentation
caused by MiFID,112 as discussed in Section III.A.1. Regulators must weigh
the benefits of cross­market consistency against the possible harms of prop-
agating erroneous or panic­induced price movements across markets before
circuit breakers, or cooler heads, have a chance to engage.113 Greater sys-
temic risk is a high price to pay for miniscule increases in consistency
between markets.

E. Manipulation and Market Integrity

Due to the speed of the trades and the complexity of the algorithms, it is
difficult for regulators to detect when HFTs engage in manipulative or ille-
gal behavior. Officials have openly admitted that they lack the tools to
effectively monitor HFT.114 Anecdotal accounts of HFTs engaging in price
manipulation abound,115 and FINRA has sanctioned at least one HFT firm
for a blatantly manipulative strategy.116 Whether justified or not, the fear of
manipulative HFT has driven some retail investors out of the market.117
Regulators are accordingly concerned about rooting out market abuse and
reassuring market participants that they are safe from exploitation by HFT.
HFTs can use their superior speed profitably to deceive other market
participants. Three established abusive strategies carry the nicknames “stuff-
ing,” “smoking,” and “spoofing.” 118 “Stuffing” involves submitting huge
numbers of orders—most of which will be cancelled prior to execution—so
that the exchange becomes congested and slow traders’ information be-
comes unreliable. HFTs then trade against misinformed orders they

111. Gomber et al., supra note 21, at 6.

112. Id. at 11.
113. See Int’l Org. for Sec. Comm’ns, supra note 37, at 29–30.
114. For example, commenting on the need for a Consolidated Audit Trail system to
monitor HFT, FINRA’s VP for market regulation noted that “there’s an expectation gap be-
tween what market participants expect and what we actually have.” James Armstrong,
Officials Call CAT ‘Long Overdue’, Traders Mag. (Sept. 21 2011), http://www.
tradersmagazine.com/news/cat-sec-finra-109439-1.html. SEC Chairman Mary Schapiro simi-
larly told Congress in March 2011 that “the SEC’s ability to collect trading data is ‘wholly
inadequate to the task of overseeing the largest equity markets in the world.’ ” Scott Patterson,
SEC Pushes Plan for Audit System, Wall St. J. (Sept. 21, 2011), http://online.wsj.com/
article/SB10001424053111904491704576574883908453622.html.
115. See, e.g., HFT Quote Stuffing Market Manipulation Caught in the Act, Zero-
Hedge.com (Aug. 25, 2011), http://www.zerohedge.com/news/hft-quote-stuffing-market-
manipulation-caught-act (market observers purporting to find evidence of quote stuffing).
116. Press Release, Fin. Indus. Reg. Auth., FINRA Sanctions Trillium Brokerage
Services, LLC, Director of Trading, Chief Compliance Officer, and Nine Traders $2.26 Mil-
lion for Illicit Equities Trading Strategy (Sept. 13, 2010), available at http://www.finra.org/
Newsroom/NewsReleases/2010/P121951
117. Int’l Org. for Sec. Comm’ns, supra note 37.
118. Biais & Woolley, supra note 4, at 8–9.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

148 Michigan Telecommunications and Technology Law Review [Vol. 19:131

induce.119 In “smoking,” HFTs post generously priced limit orders with the
intention of inducing a flow of slow marketable orders. The HFTs then can-
cel their generously priced limit orders before they execute and trade with
the incoming marketable orders on more advantageous terms.120 In “spoof-
ing,” HFTs place large limit orders to sell that are above the best asking
price, with the intention of quickly cancelling them if the price moves up-
wards so that they will not be executed. The HFTs hope during spoofing that
the size of the sell orders will scare other traders into selling at a low price,
thus allowing the HFTs to scoop up a bargain.121 This list of deceptive strat-
egies is not exhaustive, but it provides insight into HFTs’ ability to use
speed to illegally hoodwink slower investors.122

III. Regulatory Responses

Responding to the concerns outlined above, regulators and exchanges
have taken a range of measures to control HFT. Yet many promising ideas
have not yet been implemented, and consensus on some points remains elu-
sive. Regulators have taken serious steps to avoid other flash crashes, but a
comprehensive regulatory approach capable of addressing all of HFT’s is-
sues has yet to emerge. A lack of conclusive evidence on certain questions is
no doubt partially responsible,123 as is financial regulators’ preoccupation
with other pressing issues during the last few years. These factors aside, we
now know enough about HFT that global regulators should feel comfortable
acting aggressively. Regulators must move beyond the limited paradigm of
protecting against flash crashes, and instead acknowledge that the benefits of
widespread, unmonitored HFT do not justify the costs and risks it imposes.
This discussion will first draw a distinction between three complemen-
tary, non-mutually exclusive regulatory perspectives on HFT. The first
perspective, which emerged in response to the Flash Crash, sees HFT as a
potential source of systemic risks. The second perspective approaches HFT
as a potential location of illegal or deceptive activity. The third perspective
looks at HFT as a potentially harmful influence on day­to­day market func-
tioning. Each of these perspectives reflects distinct, non-mutually exclusive
empirical conclusions about HFT’s effects and urges different remedies.
Next, this Note will contend that all three of these perspectives deserve
a place in the U.S. and European regulatory schemes. It will also catalog the
implications of the current state of affairs, in which regulators lack adequate

119. Id.
120. Id.
121. Id.
122. See infra Part III.A.2 (discussing the legal implications of HFT market manipula-
tion).
123. For example, compare Zhang, supra note 54, with Brogaard et al., supra note 91,
for a discussion on price discovery.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 149

visibility into HFT activity.124 If regulators cannot monitor and understand

HFT activity in detail, they cannot enforce targeted rules that exclusively
affect “bad apple” HFT strategies.125 To the extent that detailed oversight is
impracticable, broad­brush (and occasionally overinclusive) rules on HFT
may be appropriate. Finally, since the existing evidence does not demon-
strate that a slowdown in HFT would substantially harm the markets,
regulators should not shy away from aggressive measures.

A. Three Complementary Perspectives for Regulators

In the previous Section, this Note discussed five aspects of market func-
tioning that may be impacted by HFT: liquidity, volatility, price discovery,
market resiliency, and market integrity. These aspects of market functioning
interact in complex ways. For example, if HFT removes liquidity during
times of market unpredictability, it may exacerbate volatility.126 On the other
hand, if it provides liquidity under normal circumstances, it may facilitate
price discovery—yet this effect may be mitigated if it provides liquidity se-
lectively to uninformed counterparties. 127 This makes for an extremely
knotty and nuanced range of possible views on HFT. However, this tangled
range of views can be rendered manageable. We can summarize regulators’
views on HFT by examining how they answer the following three questions:
1. Is HFT a systemic risk? Regulators on both sides of the Atlantic
largely answer affirmatively in light of the Flash Crash and have
taken precautionary measures.128
2. Is HFT a likely locus of illegal market manipulation? Regulators
on both sides of the Atlantic acknowledge that this is a potential-
ly large problem, and they currently lack the tools and resources
to deal with it effectively.129
3. Does HFT harm market quality (i.e., liquidity, volatility, and
price discovery) on a day­to­day basis, regularly imposing

124. See Brogaard et al., supra note 91.

125. “Targeted rules” refers to those rules that target subsets of HFT, i.e., rules aimed
specifically at market manipulators or at HFT algorithms engaging in particular statistical
arbitrage strategies. “Broad­brush regulation” means measures that do not depend on the abil-
ity to distinguish between different HFT strategies, like circuit breakers or across-the-board
transaction taxes.
126. CFTC-SEC Findings, supra note 5 (finding that when HFT removes liquidity
during times of market unpredictability, it may exacerbate volatility and could potentially lead
to a flash crash).
127. See supra note 91.
128. See generally Bank for Int’l Settlements, supra note 25; Int’l Org. for Sec.
Comm’ns, supra note 37.
129. See, e.g., Haldane, supra note 14.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

150 Michigan Telecommunications and Technology Law Review [Vol. 19:131

serious negative externalities? This is a more controversial prop-

osition, with many regulators apparently undecided.130
The regulatory perspectives that flow from affirmative answers to these
questions are not mutually exclusive. In other words, there is no contradic-
tion in seeing HFT as a potential systemic risk, a likely locus of illegality,
and a detriment to everyday market quality. In general, regulators give a
stronger affirmative answer to the first question than to the second two. This
is understandable given that the Flash Crash armed regulators with tangible
evidence of HFT’s contribution to systemic risk. On the other hand, the evi-
dence also compels a strongly affirmative answer to the second question and
a weakly affirmative or neutral answer to the third question. Adopting all
three of these perspectives simultaneously is key to developing an efficient
and comprehensive policy on HFT, as illustrated by Figure 1 below.

Figure 1
1. Systemic Risk 2. Illegality 3. Harm to Day-to-Day
Market Quality
Regulatory remedy • Circuit breakers • Consolidated Audit • Pigovian tax schemes
urged • Enforceable market- Trail • Limit up/limit down rules
making obligations • Increased securities • Resting rules
• Naked-access ban law enforcement
• Regulatory review of
algorithms
Regulatory remedies If HFT imposes systemic risks, hides illegal activity, and has a negative or neutral
urged by the three effect on market quality, its harms likely outweigh its benefits. The three propositions
perspectives on HFT taken together therefore suggest that more restrictive regulatory measures—like
risk, when taken resting rules131 and cancellation or transaction taxes—are probably justified.
together

As a whole, U.S. and European regulators have taken the threat of sys-
temic events like the Flash Crash seriously, placing them on the right track
regarding HFT policy. Some abusive HFT methodologies—most notably
naked or unfiltered access—have been eliminated.132 In November 2010, the
SEC approved rules banning stub quotes—posted orders at grossly implau-
sible prices that can exacerbate price swings.133 Furthermore, in June 2012,
the SEC announced its approval of limit up/limit down rules.134 U.S. ex-

130 See, e.g., Bank of Int’l Settlements, supra note 25; Int’l Org. of Sec.
Comm’ns, supra note 37.
131. Haldane, supra note 14, at 17–18.
132. Press Release, SEC, SEC to Publish for Public Comment Updated Market-Wide
Circuit Breaker Proposals to Address Extraordinary Market Volatility (Sept. 27, 2011), avail-
able at http://www.sec.gov/news/press/2011/2011-190.htm.
133. Press Release, SEC, SEC Approves New Rules Prohibiting Market Maker Stub
Quotes (Nov. 8, 2010), available at http://www.sec.gov/news/press/2010/2010-216.htm.
134. Press Release, SEC, SEC Approves Proposals to Address Extraordinary Volatility in
Individual Stocks and Broader Stock Market (June 1, 2012), available at
http://www.sec.gov/news/press/2012/2012-107.htm; see also Joint CFTC-SEC Advisory
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 151

change rules are also moving in the right direction: most recently, Nasdaq
and Direct Edge implemented fees, albeit on fairly lenient terms, for very
large order cancellations.135 European regulators, too, are discussing serious
changes in HFT regulation for MiFID II, which could take effect around
2015.136 But risks remain, and far too little has been done to root out HFT
market manipulation. Although regulators have discussed good ideas,
change has been tentative and incremental.
While caution regarding unintended consequences is warranted,137 def-
erence to the questionable evidence of HFT’s market benefits is not. The
empirical research does not demonstrate that HFT has enough clear social
utility to justify its clear risks. Below, this Note explains and evaluates re-
cent regulatory actions within the three perspectives of systemic risk,
illegality, and market quality. It builds toward the conclusion that regulators
should move forward confidently with measures like resting rules or cancel-
lation taxes that would broadly alter the practice of HFT.

1. High-Frequency Trading as a Systemic Risk

The Flash Crash served as a wake­up call that alerted many regulators to
the threat of HFT-related systemic crashes.138 Consequently, regulators have
taken a proactive stance on this aspect of HFT. In 2011, the CFTC and SEC
proposed a basket of rational safeguards against other flash crashes, and pur-
sued regulations implementing them. 139 These included updated circuit
breakers, limit up/limit down mechanisms,140 stub quote bans, and naked-
access bans.141 The SEC has put some of these proposals into law, updating

Comm. on Emerging Regulatory Issues, Recommendations Regarding Regulatory

Responses to the Market Events of May 6, 2010 (2011), available at
http://www.cftc.gov/ucm/groups/public/@aboutcftc/documents/file/jacreport_021811.pdf
[Hereinafter Joint Advisory Recommendations].
135. Telis Demos, U.S. Bourses to Fine HFT Data-Cloggers, Fin. Times (Mar. 7, 2012,
11:55 PM), http://www.ft.com/intl/cms/s/0/8d3aead4-689b-11e1-b803-00144feabdc0.html#
axzz1q4odaTL6.
136. See generally Oliver Linton, Maureen O’Hara & Jean-Pierre Zigrand, Economic
Impact Assessments on MiFID II Policy Measures Related to Computer Trading in Financial
Markets (U.K. Gov’t Office of Sci., Foresight Project Working Paper, Aug. 2012).
137. The main worry is diminishing market liquidity, discussed in Section II.A. Another
worry is that overzealous regulation of algorithms’ content may cause a proliferation of simi-
lar algorithms that will fail to offset one another in the market, thus creating systemic risks.
See Gomber et. al, supra note 21, at 60–61. This concern seems primarily directed at detailed
regulation of algorithms’ content (as opposed to broad-brush measures like mandatory resting
periods), but more research is warranted.
138. See, e.g., CFTC-SEC Findings, supra note 5, at 6–8; Bank of Int’l Settle-
ments, supra note 25, at 15–17; Int’l Org. for Sec. Comm’ns, supra note 37, at 11–12.
139. See Joint Advisory Recommendations, supra note 134.
140. Limit up/limit down mechanisms restrict the speed at which the price of an ex-
change-traded security can fluctuate. See id. at 5; see also Haldane, supra note 14, at 17–19
(discussing the proposed regulations in the U.S. and Europe).
141. See Joint Advisory Recommendations, supra note 134, at 4.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

152 Michigan Telecommunications and Technology Law Review [Vol. 19:131

the circuit-breaking rules several times, and most recently supplementing

them with a limit up/limit down regime.142 Additional regulatory achieve-
ments include eliminating stub quotes and naked access in U.S. markets.143
SEC Chairman Mary Schapiro has also claimed that the updated circuit-
breaking rules prevented Knight Capital’s rogue algorithm from causing
greater damage.144 Still tighter SEC circuit-breaking rules will go into effect
in February 2013, at the same time as the limit up/limit down regime.145
Regulators have further discussed imposing market-making guidelines on
HFTs, but continue to allow exchanges to manage these obligations.146 To-
gether, these measures should reduce the risk of a Flash Crash repeat. But
how do they work?
Circuit breakers are mechanisms designed to prevent or correct anoma-
lous trades or halt trading when there is evidence of dangerous volatility.147
SEC-mandated circuit breakers have been in place since 1988, but prior to
2012, they were only triggered once, in 1997.148 Unhappily, they failed to
trigger during the Flash Crash because “the downturn was not broad
enough.”149 Accordingly, the SEC implemented a new, more sensitive sys-
tem in September 2010.150 These new circuit breakers do not require a broad
downturn to trigger; rather, they temporarily stop trading in systemically
important individual stocks and ETFs if their prices move rapidly within a
five­minute period.151 This should reduce the severity of irrational, non-

142. Notice of Filing of a National Market System Plan to Address Extraordinary Market
Volatility, 76 Fed. Reg. 31647-01 (June 1, 2011); Press Release, SEC, SEC Announces Filing
of Limit Up-Limit Down Proposal to Address Extraordinary Market Volatility (Apr. 5, 2011),
available at http://www.sec.gov/news/press/2011/2011-84.htm.
143. Press Release, SEC, SEC to Publish for Public Comment Updated Market-Wide
Circuit Breaker Proposals to Address Extraordinary Market Volatility (Sept. 27, 2011), availa-
ble at http://www.sec.gov/news/press/2011/2011-190.htm.
144. Press Release, SEC, Chairman Schapiro Statement on Knight Capital Group Trad-
ing Issue (Aug. 3, 2012), available at http://www.sec.gov/news/press/2012/2012-151.htm.
145. Investor Bulletin: New Measures to Address Market Volatility, SEC (July 23,
2012), http://www.sec.gov/investor/alerts/circuitbreakersbulletin.htm.
146. Market-making guidelines are obligations designed to prevent liquidity providers
from fleeing the market in times of stress. These would mirror the liquidity-providing obliga-
tions of exchange “specialists”—analog market makers who have largely been supplanted by
HFT. Joint Advisory Recommendations, supra note 134, at 10.
147. CFTC-SEC Findings, supra note 5, at 7.
148. Press Release, SEC, supra note 143.
149. Investor Bulletin: New Stock-by-Stock Circuit Breakers, SEC (Aug. 9, 2011),
available at http://www.sec.gov/investor/alerts/circuitbreakers.htm.
150. Press Release, SEC, SEC Approves Rules Expanding Stock-by-Stock Circuit
Breakers and Clarifying Process for Breaking Erroneous Trades (Sept. 10, 2010), available at
http://www.sec.gov/news/press/2010/2010-167.htm.
151. Investor Bulletin, supra note 149; SEC Press Release, supra note 150, at 2 (“For
stocks priced $25 or less, trades will be broken if the trades are at least 10 percent away from
the circuit breaker trigger price. For stocks priced more than $25 to $50, trades will be broken
if they are 5 percent away from the circuit breaker trigger price. For stocks priced more than
$50, the trades will be broken if they are 3 percent away from the circuit breaker trigger
price.”).
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 153

fundamental shocks. The 2011 “stock­specific” circuit breakers have indeed

mitigated (though not prevented) harm by stopping trading in stocks affect-
ed by Knight Capital’s rogue algorithm on August 1, 2012.152 In addition to
these stock­specific circuit breakers, the SEC has approved measures mak-
ing market­wide circuit breakers easier to trigger, which will go into effect
on February 4, 2013.153
The SEC goes beyond circuit breakers in its attempt to control HFT-
related systemic risk. It also restricts stub quotes in equity markets.154 Stub
quotes, once again, are offers to buy or sell at prices unrealistically far from
the current market price.155 Stub quotes played a key role in the Flash Crash
because when other liquidity providers fled the market, HFTs actually con-
summated transactions at wildly low prices, causing the market to plunge.156
The stub quote ban is a prudent measure that should reduce the risk of a
Flash Crash recurrence. Furthermore, the SEC has adopted measures clari-
fying the rules on erroneous trade nullification, requiring broker dealers to
control risk within private trading pools, and prohibiting brokers from giv-
ing clients naked access to public exchanges.157 The naked access ban closes
a dangerous loophole that allowed HFTs to get direct access to exchanges
by trading on their broker’s account, thus avoiding risk checks and capital
requirements.158 Naked access therefore compounded HFT’s other problems
by adding counterparty risk.
These rules are not perfect and do not reach all non-equity markets
where HFT is also abundant (e.g., FX markets). Intermarket cross-linkages
remain extremely tight due to rapid HFT arbitrage, so that volatility in one
market sector could threaten wider markets. Overall, however, the SEC has
been fairly proactive and may deserve credit for the fact that the Knight
Capital algorithm did not cause disastrous cross­market propagation.
Additional important safeguards, including a limit up/limit down mech-
anism restricting single­stock short­term volatility, will take effect in 2013,
further protecting U.S. markets.159 By preventing the most disruptive and
irrational trades from occurring in the first place (rather than stopping trad-
ing after they occur), this measure will likely improve markets’ resiliency.160

152. Press Release, SEC, supra note 144.

153. Investor Bulletin: New Measures to Address Market Volatility, SEC (July 23,
2012), http://www.sec.gov/investor/alerts/circuitbreakersbulletin.htm.
154. Quotation Standards for Market Makers, Exchange Act Release No. 34-63255,
2010 WL 4466998 (Nov. 5, 2010), available at http://www.sec.gov/rules/sro/bats/2010/34-
63255.pdf.
155. Id.
156. CFTC-SEC Findings, supra note 5, at 38–39.
157. Id.
158. Gomber et al., supra note 21, at 41.
159. Investor Bulletin, supra note 153.
160. Id. (“Because single-stock circuit breakers are triggered after a trade occurs at or
outside of the applicable percentage threshold, circuit breakers have been triggered by errone-
ous trades. In contrast, the new limit up-limit down mechanism is intended to prevent trades in
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

154 Michigan Telecommunications and Technology Law Review [Vol. 19:131

In sum, U.S. regulators have probably reduced the likelihood of a systemic,

Flash Crash–like recurrence, but risks remain.
European authorities have been relatively slow to implement similar
measures,161 although European exchanges have long had stock-specific cir-
cuit breakers.162 Nevertheless, European lawmakers are currently debating a
range of aggressive proposals for the new iteration of the MiFID. Proposed
versions of the legislation would mirror many actions taken by U.S. regula-
tors, and in some cases go much further in restricting HFT.
In a July 2011 speech, the Bank of England’s Executive Director of
Financial Stability argued for three aggressive measures dealing with HFT-
related crashes: stricter market-making guidelines, strict circuit breakers,
and resting rules.163 Market-making guidelines would obligate HFTs to con-
tinue providing liquidity under adverse market conditions.164 Many criticize
this idea because under stressful market circumstances, market makers may
prefer to incur fees than to risk disastrous trades;165 but if implemented sen-
sitively, it has the potential to mitigate the effect of HFTs withdrawing
liquidity, as they did in the Flash Crash. Resting rules, explained further
below, would impose a minimum trading speed and thus reduce the speed at
which market makers could withdraw liquidity.166
Other voices in Europe disagree with these aggressive proposals. In a
paper commissioned by the Deutsche Börse Group, Gomber et al. argue that
the Flash Crash was exclusively a U.S. problem, rooted largely in the ab-
sence of stock-specific circuit breakers on May 6, 2010.167 They further
argue that HFT has an intermarket arbitrage role that is more important in
the U.S. than in Europe.168 This argument relies on the fact that while both
jurisdictions have recently seen the proliferation of smaller exchanges and
trading venues, Europe lacks the equivalent of Regulation NMS. Regulation
NMS requires U.S. orders to be routed to the exchange offering the “nation-
al best bid or offer.”169 Therefore, the argument goes, U.S. traders can trust
that they will get the best price available, but European traders need tighter
HFT arbitrage to be confident that an order executed on any given exchange

individual securities from occurring outside of a specified price band. These price limit bands
will be 5%, 10% or 20%, or the lesser of $0.15 or 75%, depending on the price of the stock.
Additionally, these price bands will double during the opening and closing periods of the
trading day. If the stock’s price does not naturally move back within the price bands within 15
seconds, there will be a five-minute trading pause.”)
161. Haldane, supra note 14, at 17.
162. Gomber et al., supra note 21, at 48.
163. Haldane, supra note 14, at 17–19.
164. See supra note 122.
165. See Gomber et al., supra note 21, at 53.
166. See Haldane, supra note 14.
167. Regulation NMS requires brokers to execute market orders on the market with the
best available price, thus increasing intermarket linkages. Gomber et al., supra note 21, at 1.
168. Id. at 11–12.
169. See Gomber et al., supra note 21.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 155

will get the best possible price. This argument stakes too much on the effi-
cacy of stock-specific circuit breakers, and likely overestimates the
importance of extremely short-term price arbitrage.
While the legislative process continues, the draft version of MiFID II
includes changes that will affect HFT. If it is adopted, HFTs will no longer
be able to escape the Directive’s requirements through an exemption in Mi-
FID Article 2.170 That exemption, intended for “persons who do not provide
any investment services or activities other than dealing on their own ac-
count,” will no longer apply to HFTs.171 Thus, HFTs will need to comply
with basic reporting requirements and rules on internal risk controls.172 Fur-
thermore, changes to MiFID Article 17 may require HFTs to disclose details
of their algorithms to regulators, to ensure that they provide liquidity irre-
spective of market conditions.173 It is not yet clear how this potentially
sweeping provision would be enforced or whether it would apply to all
HFTs.174
Amidst industry opposition, a few aggressive proposals discussed in the
MiFID review process—such as amendments to MiFID Articles 14 and 39
that would have imposed liquidity provision obligations and limited the
speed at which HFTs could cancel orders175—did not make it into the Octo-
ber 2011 draft of the proposed directive.176 However, a September 26, 2012
vote by the European Parliament’s economic affairs committee officially
revived the idea of a speed limit or “resting period,” which would require
traders to let limit orders remain open for a half-second before cancella-
tion.177 Such a “resting rule,” which may become law when MiFID II goes
into effect in 2014 or 2015, could mitigate liquidity droughts like the Flash
Crash and also address a range of other concerns about HFT. Perhaps most
importantly, resting rules could reduce illegal or deceptive trading, as dis-
cussed below.

2. High-Frequency Trading as a Locus of

Illegality or Deceptive Practices
Regulators currently lack the ability to effectively monitor and analyze
HFT activity. If regulators had better information, HFT market manipulation
would likely be prosecutable in both the U.S. and Europe. Indeed, tactics
such as stuffing, smoking, and spoofing fall within the commonplace under-
standing of market manipulation. Because these techniques aim to induce

170. Proposed Directive, supra note 43, at art. 2.

171. Id. at 7.
172. Id. at 7–8.
173. See Linton, O’Hara & Zigrand, supra note 136, at 17–19.
174. See id.
175. See Gomber et al., supra note 21, at 50.
176. See Proposed Directive, supra note 43.
177. Huw Jones, EU Lawmakers Vote for Sweeping Market Reforms, Reuters (Sept.
26, 2012), http://www.reuters.com/assets/print?aid=USBRE88P0BJ20120926.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

156 Michigan Telecommunications and Technology Law Review [Vol. 19:131

misinformed trading by counterparties, they most likely count as “deceptive

devices” under Rule 10b-5 of the Exchange Act.178 Under new rulemaking
powers given to it by the Dodd-Frank Act, the CFTC in 2011 promulgated
an analogous rule, 17 C.F.R. § 180.1, banning deceptive practices in the
trading of swaps, commodities, and futures.179 Manipulative HFT strategies
are also unlawful under current European law. Under Article 43 of MiFID,
exchanges must be required by Member States to identify and report “mar-
ket abuse.”180 Comments in a new draft of the Market Abuse Directive
(“MAD”) specify that the definition of market abuse in the current version
of MAD probably already encompasses “some . . . [HFT] strategies such as
quote stuffing, layering, and spoofing” and that the new draft should prohibit
these strategies even more clearly.181
Thus, regulators largely have the legal ability to prosecute market abuse
by HFTs if they can conclusively discover it; however, scienter requirements
complicate matters. For example, if an HFT submits and then cancels a large
number of limit orders, it is hard to know whether they intended to deceive
counterparties, or whether they revised their order legitimately based on new
information.182 Rule 10b-5 requires only “strong circumstantial evidence of
conscious misbehavior or recklessness,”183 and 17 C.F.R. § 180.1 similarly
prohibits deceptive behavior that is either “intentional or reckless.”184 The
recklessness standard will probably render it challenging, but not impossi-
ble, for regulators to prove HFT market abuse if they have detailed and
well-analyzed market data.
As a prerequisite to prosecuting abuse, regulators need more sophisti-
cated systems to sift through enormous amounts of trading information. One
commentator illustrated the situation colorfully: “The traders are driving
Ferraris, and the market policemen—the regulators—are riding bicycles.”185
Accordingly, the SEC is moving forward with the development of a Consol-
idated Audit Trail, a comprehensive system for merging data between
different markets and different participants.186 The SEC estimates the cost of
developing the system at $4 billion, which would be recouped through fees

178. 17 C.F.R. § 240.10b–5 (2012).

179. 17 C.F.R. § 180.1 (2012).
180. Eur. Sec. & Mkt. Auth, supra note 85, at 44.
181. Proposal for a Regulation of the European Parliament and of the Council on Insid-
er Dealing and Market Manipulation (Market Abuse), at 8, COM (2011) 651 final (Oct. 20,
2011), available at http://ec.europa.eu/internal_market/securities/abuse/index_en.htm.
182. See, e.g., Jeremy Grant, European MP Calls for Holding Period in HFT, Fin.
Times (Mar. 23, 2012), http://www.ft.com/cms/s/0/5024ede6-74f5-11e1-ab8b-00144feab49a.
html#axzz1q4odaTL6.
183. Novak v. Kasaks, 216 F.3d 300, 308 (2d Cir. 2006).
184. 17 C.F.R. § 180.1(a) (2012).
185. Bonnie Kavoussi, SEC May Monitor High­Frequency Trading With Consolidated
Audit Trail, Huffington Post (Oct. 10, 2011), http://www.huffingtonpost.com/2011/10/
10/sec-high-frequency-trading_n_987378.html; see also supra note 99.
186. CFTC-SEC Findings, supra note 5, at 14.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 157

to FINRA and the exchanges; these entities may decide to pass the cost on
to HFTs.187 MIT economist Andrew Lo comments that this system would be
a bargain if spread out over the course of several years and if it helped regu-
lators manage turmoil and restore confidence to retail investors.188 Yet it
would be a waste if, after enormous investment, regulators remained a step
behind the curve.
In September 2010, the SEC took the aggressive step of asking some
HFT firms to disclose their algorithms to the agency in order to aid in the
scrutiny of possible market manipulation.189 Although this signals serious-
ness about stopping manipulation, it is questionable whether the SEC can
usefully interpret the algorithms.190 The Office of Financial Research, a
creation of the Dodd-Frank Act meant to support the data analysis capability
of the Financial Stability Oversight Board,191 may have the capability to
perform the needed analyses. Therefore, if algorithm monitoring is to be-
come an ongoing part of the regulatory scheme, intensive and continuous
interagency cooperation might prove necessary. European regulators face
similar workability questions regarding the proposed algorithm disclosure
requirements in MiFID II.192
An exception to the apparent unenforceability of HFT market abuse
came in the case of Trillium Brokerage Services in September 2010.193
FINRA levied a $2.26 million fine against Trillium and nine of its traders
for engaging in high­frequency trades aimed at deceiving counterparties.194
However, this fine reflects an industry self-regulatory action, not a legal
sanction. FINRA’s report on the incident leaves it unclear how the violation
was detected. 195 Furthermore, FINRA has taken no comparable action
against an HFT firm since Trillium.
In short, regulators worldwide lack the ability to effectively monitor
markets for HFT abuse and reassure investors of market integrity. Better
monitoring tools like the SEC’s Consolidated Audit Trail (or the more re-
cently announced “Midas” system) may help,196 but it is far from clear that

187. Scott Patterson, SEC Pushes Plan for Audit System, Wall St. J. (Sept. 21, 2011),
http://online.wsj.com/article/SB10001424053111904491704576574883908453622.html.
188. See Kavoussi, supra note 185.
189. See Lynch & Spicer, supra note 95.
190. Id.
191. See U.S. Dep’t of the Treasury, Office of Fin. Research, Strategic Frame-
work 2 (Mar. 2012), available at http://www.treasury.gov/initiatives/wsr/ofr/Documents/
OFRStrategicFramework.pdf.
192. See Linton, O’Hara & Zigrand, supra note 136, at 17–19.
193. Press Release, Fin. Indus. Reg. Auth, supra note 116.
194. Id.
195. Letter from Trillium Brokerage Servs., LLC, et al., to FINRA, Letter of Ac-
ceptance, Waiver and Consent (Aug. 5, 2010), available at http://www.finra.org/web/groups/
industry/@ip/@enf/@ad/documents/industry/p122044.pdf.
196. See Nathaniel Popper and Ben Protess, To Regulate Rapid Traders, S.E.C. Turns To
One Of Them, N.Y. Times (Oct. 7, 2012), http://www.nytimes.com/2012/10/08/business/
sec-regulators-turn-to-high-speed-trading-firm.html?pagewanted=all.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

158 Michigan Telecommunications and Technology Law Review [Vol. 19:131

such tools will yield enough information to prove scienter in market manip-
ulation crimes. By contrast, comprehensive measures curtailing HFT order
cancellation could greatly diminish HFTs’ ability to use misdirection to de-
ceive other market participants. If such measures work as intended, they
would stop abuse before it happened, rather than scouring complex data to
find it after the fact.
European regulators have been more aggressive in this area. Markus
Ferber, German member of the European Parliament and a leader in the Mi-
FID review process, has advocated for a resting rule in stocks and
derivatives.197 On September 26, 2012, the European Parliament’s economic
affairs committee approved a measure requiring limit orders to remain valid
for at least five hundred milliseconds, placing the measure on the road to
becoming law.198 If structured carefully, this could make it difficult or im-
possible for HFTs to post limit orders that they have no intention to fulfill.199
That, in turn, could undermine strategies like stuffing, smoking, and spoof-
ing, and reduce the problem of illusory liquidity.
Andrew Haldane of the Bank of England has strongly argued for resting
rules.200 He explains the key distinction between this regulatory proposal
and others, like market­making guidelines and circuit breakers: “Minimum
resting periods are an ex-ante, non-state contingent intervention rule. They
tackle the arms race at the source by imposing a speed limit on trading.”201 A
well-designed resting rule could mitigate many of HFT’s problems, from
flash crashes to market abuse. Indeed, the most problematic HFT market
strategies require the ability to cancel orders quickly. A resting rule would
limit that possibility, forcing HFTs to assume a risk familiar to analog trad-
ers: namely, if conditions change fractions of a second after you place a
limit order, a counterparty may trade with you before you can retract. The
potential downside is that if HFT benefits market quality, minimum resting
or holding periods would reduce those benefits by increasing the cost of
HFT market making.
Therefore, regulators seeking to eliminate market abuse should closely
scrutinize the question of whether HFT’s market quality benefits substantial-
ly outweigh accompanying harms. The clear evidence of HFT’s potential for
market abuse should shift the burden to HFTs to prove overriding benefits to
market quality. If unambiguous evidence of substantial benefit to market
quality cannot be found—and the research largely suggests it cannot—
regulators ought not shy away from major rule changes like resting periods
or order cancellation taxes (described below). These measures could greatly
increase both the actual and perceived fairness of markets.

197. See Grant, supra note 182.

198. See Jones, supra note 177.
199. See supra notes 89, 90.
200. Haldane, supra note 14, at 18–19.
201. Id.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 159

3. High-Frequency Trading as a Detriment

to Day-to-Day Market Quality
Few regulators have fully embraced the position that HFT harms market
quality on a day-to-day basis. To be sure, some regulators have acknowl-
edged concerns that HFT may “adversely affect the quality of markets, for
instance, through the decrease of trade size and by pushing up indirect trad-
ing costs for retail and institutional investors.”202 It is also widely accepted
that HFT’s liquidity provision is not as beneficial as it appears because
“[t]he operational model of HFT requires trading in markets that are already
liquid enough to be able to quickly enter and exit from the market. This is a
critical requirement for limiting their exposure to market risk.”203 But these
observations have not been taken to their logical conclusion: that HFT’s
benefit to market quality is more doubtful than it appears.
Regulatory output has relied heavily on work by Hendershott and
Riordan, which seems to indicate that HFT benefits market quality.204 Yet in
their 2011 study, these researchers admit a failure to establish that HFT’s
contribution to price discovery is either lasting or reflective of information
that would not have been quickly traded on by other market participants
anyway.205 Nonetheless, regulators tend to lean toward concluding that the
effect of HFT on market quality is “neutral to beneficial.”206
This overreliance on limited and imperfect research goes some way to-
ward explaining the fact that regulators on both sides of the Atlantic have
not yet implemented stringent measures such as resting rules or steep can-
cellation fees. Resting rules would simultaneously address many concerns
about HFT, from rapid liquidity taking to market manipulation. They would
also reduce socially inefficient investments in ever-faster trading infrastruc-
tures and mitigate the disadvantage to slower traders trying to impound valid
information into stock prices.207

202. Eur. Sec. & Mkt. Auth, supra note 85, at 10.
203. Int’l Org. for Sec. Comm’ns, supra note 37, at 25.
204. Eur. Sec. & Mkt. Auth., supra note 85, at 64; Bank of Int’l Settlements, supra
note 25, at 28; Int’l Org. for Sec. Comm’ns, supra note 37, at 25.
205. See supra notes 103–104 and accompanying text.
206. The following quote concisely illustrates how many regulators have interpreted the
empirical work on HFT: “Overall, the empirical literature suggests that algorithmic trading
and HFT are neutral to beneficial for market quality, in that volatility has declined and quoted
spreads have narrowed. The findings on liquidity provision are mixed. Generally, algorithmic
liquidity appears to be more strategic than non-algorithmic liquidity and tends to decline when
volatility rises. Regarding price discovery, contrary conclusions have been drawn for FX and
for equities. Taken as a whole, the literature consistently shows that information in orders has
become more relevant than information in trades.” Bank of Int’l Settlements, supra note
25, at 13; see also Int’l Org. for Sec. Comm’ns, supra note 37, at 26–27 (“The available
evidence fails to find a consistent and significant negative effect of HFT on liquidity . . . .
[T]he limited empirical evidence available so far has not clearly identified negative effects of
HFT on the efficiency of the price discovery process.”).
207. See Biais, Foucault, & Moinas, supra note 36, at 27–28.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

160 Michigan Telecommunications and Technology Law Review [Vol. 19:131

Order cancellation fees or taxes would also address many concerns at

once, perhaps even more efficiently than a resting rule. Like a “sin” tax,
such fees would discourage traders from posting orders they do not intend to
execute. Imposing costs on cancelled orders may also diminish manipulative
HFT strategies that involve massive order cancellation—like stuffing and
spoofing—by rendering them uneconomical.208 In fact, the SEC has called
for cancellation fees,209 but they have been only incompletely self-imposed
by a couple of U.S. exchanges.210 These current fee structures aim primarily
at preventing overload in the exchange computer systems, and discourage
only the most blatantly excessive cancellations.211
Even if we accept the view that HFT’s contribution to market quality is
“neutral to beneficial,”212 these benefits are probably minor and equivocal. In
light of the clear evidence of HFT’s harms—namely, added systemic risk
and reduced market integrity—more aggressive regulation appears justified.
Yet the right structure and combination of rules on resting periods and can-
cellation or transaction fees remains a complex open question.

Conclusion
Academic analysis of HFT’s effect on markets is difficult, and further
study would help point the way forward; however, existing evidence sug-
gests a handful of general conclusions. HFT adds to systemic risk by tightly
interlinking markets and creating the possibility of Flash Crash–type events.
HFT also provides opportunities for illegal market manipulation that are
difficult and expensive to detect. On the positive side, HFT provides liquidi-
ty and narrows spreads under normal trading conditions, and its marketable
orders tend to move prices in the right direction. But empirical research has
not demonstrated that these benefits are substantial, for several reasons.
First, HFT stops providing liquidity—and even reduces it—when volatility
is high and liquidity is most in demand. Second, HFT’s apparent contribu-
tions to price discovery could be: A) an artifact of HFT’s ability to identify
and front-run informed traders, or B) the result of millisecond price arbi-
trage strategies, impounding information that the market would soon have
incorporated anyway.213
International and domestic regulators have by and large accepted these
conclusions. However, their actions until now have focused largely on
avoiding Flash Crash–like events by beefing up circuit breakers, closing
regulatory loopholes like naked access and the exemption in MiFID Article

208. See Biais & Woolley, supra note 4, at 8–9.

209. Joint Advisory Recommendations, supra note 134.
210. Demos, supra note 135.
211. See id.
212. Bank of Int’l Settlements, supra note 25, at 13.
213. See supra notes 103–104 and accompanying text.
Prewitt FTP 2M.doc 12/12/2012 8:31 AM

Fall 2012] High-Frequency Trading 161

2, and encouraging more industry self-regulation. These are steps in the

right direction, but the facts suggest that regulators should do more to ensure
market integrity. Strict order cancellation taxes or resting rules would likely
reduce systemic risk and diminish opportunities for market abuse. These
benefits would come at only a questionable cost to market liquidity and
price efficiency. Andrew Haldane, Executive Director of Financial Stability
for the Bank of England, states the current predicament articulately:
“In calibrating [the] trade-off, a judgment would need to be made
on the social value of split-second trading and liquidity provision
and whether this more than counterbalances the greater market un-
certainty it potentially engenders. At times, the efficiency of
financial markets and their systemic resilience need to be traded off.
This may be one such moment. Historically, the regulatory skew
has been heavily towards the efficiency objective. Given today’s
trading topology, it may be time for that to change.”214
Superfast trading on public exchanges is not an inevitable feature of moder-
nity, and the fact that resting rules or cancellation fees could dramatically
alter the practice of HFT does not in itself make them unreasonable. Regula-
tors should lose less sleep over diminishing the benefits of HFT, and do
more to ensure that it cannot threaten the stability and integrity of the mar-
kets.

214. Haldane, supra note 14, at 18–19.