ASIC demands “robust” controls over automated trading
The Australian financial markets regulator
has put tougher oversight of brokers’ order processing systems at the heart of
proposed guidelines to protect investors against the flash crashes and outages
that have bedevilled other equity markets.
On Monday, the Australian Securities and
Investments Commission (ASIC) published for consultation new rules and guidance
on automated trading as part of ongoing reforms to the Australian equity market
The proposed new framework, known as
Consultation Paper 184, includes rules requiring market participants to
implement filters and other controls to suspend trading, revised processes for
certifying systems that release orders into the market and new guidance on the
testing of systems and controls used to support automated trading. ASIC’s
outlined changes would require reviews of system changes “at least yearly” and
demonstration of the ability to handle highly automated trading and stress
testing of order flow.
Subject to revision following the
consultation process, the new rules will require market participants to exert
direct control over pre-trade filters and to “suspend, limit or prohibit an
order or series of orders” from an automated order processing (AOP) system in
real time. This places responsibility clearly with the market participant, i.e.
the exchange member, rather than the client. The new rules also toughen up
oversight of AOP systems by requiring a more stringent audit trail for initial
certification, implementation of material changes and annual reviews.
Noting that ASIC’s fourth report on the
supervision of Australian financial markets, also released on 13 August,
highlighted issues relating to the growth of high-frequency trading (HFT),
deputy chairman Belinda Gibson said HFT “and algorithms generally” were a cause
“Recent events overseas are a reminder of
the speed and automation of markets and the importance of robust controls over
those systems. The enhancements to participant-level controls … will build
confidence in the integrity and cleanliness of our capital markets and
facilitate international capital markets,” said Gibson.
On 1 August, a problem with a system
upgrade caused US market maker Knight Capital to send duplicate orders to the
New York Stock Exchange, resulting in a US$440 million loss for the firm. The
glitch, which follows technology-related problems earlier this year with the
listings of social networking site Facebook and exchange operator BATS Global Markets, comes despite measures implemented by US regulators after the 6 May
2010 flash crash, including tighter pre-trade controls, market-wide circuit
breakers and the introduction of a consolidated audit trail.
ASIC’s proposals are open to consultation
until 14 September with market integrity rules and regulatory guidance expected
to be released in October. These participant-level rules on automated trading
follow draft market level controls released in June as part of the regulator’s
Consultation Paper 179, ‘Australian market structure: Draft market integrity
rules and guidance’. The paper outlines ASIC’s approach on market operator
systems and controls, extreme price movements, enhanced data for market
supervision and pre-trade transparency. The consultation closed on 6 August and
final rules are slated for release in October.
“We expect investors, issuers, market participants
and market operators to benefit from greater protection against the risk of
aberrant automated trading disrupting Australia’s equity markets by raising
expectations for testing of AOP systems filters and controls,” said ASIC. The
watchdog acknowledged that market participants which do not have the necessary
controls in place would “incur some cost in building the capability” and that
the cost of system and process changes “may be material”.
Liam Madden, agency broker Instinet’s head of compliance for Asia, welcomed
the consultation but also stressed the importance of defining terms of
reference. “Benchmark algorithms (such
as VWAP and volume participation algos) are totally different to ‘black-box’
algorithms used by high-frequency traders,” he said. “It is only once these
definitions are clearly laid out that we can really hope to create an
environment conducive to constructive comment and effective regulation in this