CFTC makes a renewed effort to curb what some consider to be the excesses of automated trading.
The Commodity Futures Trading Commission (CFTC) has
released a “concept paper” aimed at continuing and
defining the regulatory response to the increased level of
automated trading. Whilst the headlines surrounding the paper are
inevitably about high frequency trading activities, the broader
intention of the paper is to discern the appropriate level of
controls at all stages of the dealing process, including trading
venue and the use of algorithms by non-HFT firms.
While many see the paper as targeting HFT, in reality it is
about automated trading, which represents a broad spectrum, of
which HFT is merely one part. Indeed, the paper makes the point
that most of the instances of market dislocation or failure have
been at exchange or trading platform level. That said, there are
suggestions that the infrastructure that is failing, is doing so as a
result of stress placed upon the infrastructure by the sheer level of
message traffic generated by HFT.
The paper studies several instances of market dislocations,
including – inevitably – the Flash Crash, but as has been
previously proven, the latter was caused by a non-HFT player
using a poorly designed algorithm in a poor fashion.
Although very few, if any, instances of market dislocation have
been directly related to a breakdown at an HFT firm, the paper
seems to represent a desire on the part of the CFTC to slow
things down a little, mainly because the Commission is focusing
on operational risk associated with increased automated trading.
The use of Kill Switches is discussed, as well price collars,
throttle controls and the need to have the same risk controls at all
touch points in the trading process – executing dealer, exchange
The idea of having a unified operational risk control framework
seems sensible at first sight, if for no other reason than everyone
will know what standards they have to adhere to. There are concerns that too rigid a framework could stifle competition and
therefore innovation, but that ignores the premise of the paper,
which is to make markets safer, rather than “better”.
The Kill Switch idea is nothing new to FX of course, however
even in FX, debate continues over where exactly the Kill Switch
should be located. The paper suggests situating the Switch at
multiple levels of the trade cycle, however this could raise the
odd legal or moral dilemma. There would have to be very strict
rules around what could trigger the Kill Switch to ensure it is
only used appropriately, but if that is the case a fair question
could be is multiple deployment a needless replication?
Another suggestion to curb runaway algos is the concept of
“price collars”, under which Kill Switches will automatically be
activated if a price is entered too far from the last trade, or the
market moves beyond certain parameters. The real challenge with
price collars is likely to be understanding how far from the
market the collars should be set. Inevitably, one feels that at some
stage a major event will take place triggering a genuine market
move of some substance, and a lot of orders will fail to be
transmitted because the price collar is in operation. Clearly this
technology can be calibrated with volatility, but it would still be
reactive and unlikely to pick up on genuine market gaps.
A serious challenge to the concept is how to create a real time
collar and be able to calibrate this quickly enough to ensure the
smooth operation of a market. Perhaps a variation on a theme is
worth considering in the form of a standardised and globally
agreed framework for the re-booking or cancellation of trades that
trade “off market”. It seems easier to undo an erroneous trade or
trades, than go through the morass of legal actions inevitable after
trades were not sent or executed during a volatile period of trading.
Another concept explored in the paper is throttle controls.
These obviously also exist in FX in the form of latency floors and green rooms although it should be noted that it is early days
and therefore the jury is still out on their success or
otherwise. The CFTC is proposing to impose throttle controls
on the execution side of the equation rather than market
making; however, there are concerns that this will not solve
the issue because it may push more HFT players into market
making and create a liquidity mirage.
The huge risk in this debate is that the outcome ultimately
and adversely impacts the market infrastructure. Like it or
not, automated trading is how business is done now. There
may be a human making a price and executing the trade, but
increasingly they are using technology. The trend is clear and
irreversible and there are concerns at the sub-theme of the
paper that automated trading needs to be tightly controlled.
It could be that the approach needs to be bifurcated with,
on one hand, efforts concentrated on making sure the market
infrastructure and risk controls are adequate to support trading
in the automated era and another investigation, which should
be separate, studying the different types of high frequency
behaviour and how to curb some of the excesses. In other
words, separate the emotive issue of HFT from the broader
issue of creating safer, more robust markets for non-HFT
players using fast technology.
The problem with the CFTC paper and indeed the process,
could be that it has not separated the two issues and any
time HFT is involved it will, in the current environment,
grab the headlines and take away from the other, equally
critical element of the debate. Typically HFT firms have
their operational risk nailed down, and it is not the hardest
thing to do because they are much smaller than the global
brokerages and banks. CFTC could be better directed to
stamp out bad practice rather than regulating the entire
segment out of existence.
If HFT is merely seen as just another sector that will
have to adhere to the rules, then the industry is able to
focus on the important issues that are ensuring the platform
providers have technology that is sufficiently robust and
scalable, and that those traders executing larger orders
know what they are doing and have the right tools and
services at their disposal.
Rather than focus on HFT and its supposed power to
destroy markets, it is good that the paper, in its broad remit
looks at the robustness of market venues’ technology and,
equally importantly, how non-HFT players use algorithms.
This process could bring benefits for the FX industry.
The execution algos are likely to be a tremendously
powerful tool in the coming years as more clients seek to
use them to mask activity and improve execution quality.
The industry continues to have difficulties in proving to
non-market professionals – typically treasurers, trustees and
fiduciary boards – the value and safeness of algos. It could
therefore help all in the industry if these people could be
convinced of the safety of the algo by having it adhere to