Ghost in the machine

[scandal 951]

Ghost in the machine

By Jeremy Grant and Michael Mackenzie

FT.com

Published: February 18 2010 02:00 | Last updated: February 18 2010 02:00

Not long after lunchtime one day on the New York Stock Exchange three years ago, unusual

things started to happen. Hundreds of thousands of "buy" and "sell" messages began flooding in,

signalling for orders to be made and simultaneously cancelled.

The volume of messages sent in was so large that the traffic coming into the NYSE from

thousands of other trading firms slowed, acting as a drag on the trading of 975 shares on the

board.

The case was made public only last month when the disciplinary board of the NYSE fined Credit

Suisse for failing adequately to supervise an "algorithm" developed and run by its proprietary

trading arm - the desk that trades using the bank's own money rather than clients' funds.

Algorithms have become a common feature of trading, not only in shares but in derivatives such

as options and futures. Essentially software programs, they decide when, how and where to trade

certain financial instruments without the need for any human intervention. But in the Credit Suisse

case the NYSE found that the incoming messages referred to orders that, although previously

generated by the algorithm, were never actually sent "due to an unforeseen programming issue".

It was a close call for the NYSE. Asked if the exchange could have been shut down as it was

bombarded with false trades, an exchange official says: "If you had multiplied this many times

you'd have had a problem on your hands."

Erroneous trades are not new to markets. "Fat finger" errors - mistyped orders, such as the

instance this week in which a trader mistakenly bought shares in the Japanese recruitment

company known as J-Com rather than JCom, the cable television group - are often blamed for

losses. Technological innovation is not new either. The pit trader who had a computer while rivals

still used telephones had an early advantagein the 1980s. But many blamed computers for

exacerbating the stock market crash of 1987.

Advances in technology have been so great in the past five years that markets are now

overwhelmingly driven by machines rather than humans punching orders into a keyboard. The

nightmare scenario of an exchange being knocked out by algorithms running amok, and thus

causing upheaval in the wider financial system, is seen as a real risk by many in the industry.

A decision to trade can be triggered by a news event; there is even separate technology that

"scrubs" news articles to give algorithms a sense of where - "directionally", in the jargon - it may

be profitable to start trading in a company's shares. Other types of algorithm seek out where,

across a range of exchanges and trading platforms, the best price may be found.

The speed at which such trading takes place is causing alarm. The technology is so

sophisticated that thousands of orders can be sent to an exchange's "matching engine", where

orders to buy and sell are put together, and a match found, all in less than 300 microseconds -

1,000 times faster than the blink of a human eye.

Joe Ratterman, chairman of BATS Global Markets, which operates equity exchanges in the US,

says: "The very nature of speed involves new risks, there is no way around it; it is what it is."

At the same time, markets have come to be dominated by "highfrequency traders" who rely on the

perfect marriage of technology and speed. They use algorithms to trade at ultra-fast speeds,

seeking to profit from fleeting opportunities presented by minute price changes in markets.

According to Tabb Group, a consultancy, algorithmic and high-frequency trading accounts for

more than 60 per cent of activity in US equity markets.

The transformational extent to which markets are being moved by machines, and the scale of

involvement by high-frequency firms, are raising two concerns. First, has technology reached the

point where machines pose systemic risks if they go berserk? Second, if business is now

dominated by a few participants that have this technology, does this threaten the integrity of the

markets, where a broad mixture of traders has long cohabited peacefully?

In the light of the growth of high-frequency trading, the Securities and Exchange Commission, the

US market regulator, has launched a comprehensive review of market structures. In a document

published last month, it poses questions such as: "[Does] the high speed and enormous

message traffic of automated trading systems threaten the integrity of trading centre operations?"

The Federal Reserve Bank of Chicago, part of America's central banking system, in a paper

published this month, says: "The high-frequency trading environment has the potential to

generate errors and losses at a speed and magnitude far greater than in a floor or screen-based

trading environment. Although algorithmic trading errors have occurred, we likely have not yet

seen the full breadth, magnitude and speed with which they can be generated."

John Jacobs of Lime Brokerage, which caters to what it says are "even the most complex,

automated and high-volume electronic trading strategies", says "algo" errors have certainly

occurred. In a letter to the SEC last June, he warned: "Given the growth and nature of new highfrequency

trading participants, the potential for trading-induced multiple domino bankruptcies

exists."

In his letter, he identified a number of errors, such as Morgan Stanley submitting a $10.8bn order

instead of a $10.8m order in September 2004 and a $31bn order placed by UBS in February

2009, which was 100,000 times larger than intended.

Much of the concern centres on an explosion in the amount of message traffic - electronic signals

that contain buy and sell orders - and data generated in the markets, not only by high-frequency

traders but by others including asset managers and banks, also big users of algorithms.

Stock exchanges have upgraded their capacity to absorb the rise in message traffic but questions

remain over whether they have done enough. In November the London Stock Ex-change's trading

system was knocked out for three hours after what it said were "connectivity issues" that had

affected two of the "gateways" into the exchange's order books that are used by traders. It has not

said whether a sudden surge in messages was the cause, but traders in London suspect so. It is

in the process of switching to a new technology platform that will be ready by September.

T he problem is not restricted to stock exchanges. Gerald Hanweck is chief executive of Hanweck

Associates, a risk management company bought this month by the International Securities

Exchange, a US options exchange owned by Deutsche Börse. He says: "As options trading

volume and quote traffic continues to surge, conventional market data systems are struggling to

keep pace."

NYSE Euronext, parent of the New York exchange, and other exchanges have "throttling" systems

in place that detect and prevent order message traffic exceeding acceptable levels.

Mr Ratterman says: "It's absolutely a market centre's obligation to put in place controls that

mitigate risks." BATS has audible and visible alerts that tell staff when its "ports" - gateways

through which orders arrive - are experiencing traffic of more than 3,000 messages per second.

Throttling starts at 5,000 messages per second. Asked how often that is needed, Mr Ratterman

says: "Not very often."

But Doug Rivelli, CEO of Pragma Securities, a trading technology company, echoes the view of

many market participants who say high-frequency traders have provided much needed liquidity to

the markets, which outweighs any concerns over technology going berserk. "The focus needs to

be on making sure there are pre-trade checks in place so that [the industry] can follow and catch

errors before they happen," he says.

The difficulty is that responsibility for risk controls does not lie entirely with exchanges and trading

platforms. Much of it rests instead with brokers, which increasingly provide access to such venues

under an arrangement known as "sponsored access" whereby any trading firm that is not a

member of an exchange can "piggyback" on a broker's membership to gain direct access to an

exchange. Until recently, before the SEC clamped down on the practice, traders were able to use

a form of this process - "naked access" - to gain access to exchanges without brokers conducting

pre-trade risk checks to ensure their algorithms were functioning properly.

Brokers are also responsible for "controlling the flow" of orders sent in through a sponsored

access arrangement and, where needed, choking off trades. But given that each trader may be

connected to a plethora of venues under "co-location" arrangements (see box) - each with slightly

different trading infrastructures - doing so is not always easy, according to Valerie Bannert-

Thurner at FTEN, a company specialising in pre-trade risk management.

"If you have different systems for all of these, you need to tie them all together - and in a set-up

where you have multiple co-location systems it's an integration challenge. There's no

standardisation: if all the exchanges had one simple mechanism it would be easy," she says.

Regulators are now faced with the question of how to respond. David Wright, dep uty directorgeneral

of the European Commission's internal market and services unit, says it is "a dangerous

approach to start out saying you regulate technology". He suggests that regulators should instead

pay attention to whether a particular technology "threatens market integrity and fair competition".

Jamil Nazarali, head of electronic trading at US-based broker Knight Capital, proposes that the

barriers to entry for high-frequency trading should be raised, saying firms need to have minimal

capital requirements. "The market needs better protection in a smart way, so we do not lose the

benefits of narrower bid-offer spreads and greater liquidity."

Sang Lee, managing partner at Aite Group, a US-based consultancy, says it might be possible to

"impose behavioural restrictions", such as a minimum trading speed requirement, or a

requirement that market centres "make an order valid for a certain amount of milliseconds".

He adds: "The core argument for regulating the industry is that some players with deep pockets

have a technological advantage. The dilemma is: do we slow down the faster guys or require that

the rest of the market speeds up?"

[Hilarious Clinton 452]

interesting

[thongd4me 69]

Stockbroker Fraud Blog

[M.O 6]

hmmm interesting...