If you’re driving a Porsche on the Autobahn, make sure you’ve got the right brakes
By James Williams – Last year witnessed an anticipated record high in quantitative (algorithmic) fund launches. As reported by the Financial Times recently, citing data provider Preqin, some 187 “quant” funds launched in 2011. Through the first six months of 2012, that number had already reached 95; the final figure is not yet known as its data normally has a six-month lag but Preqin’s Head of Hedge Funds, Amy Bensted was quoted as saying “We expect 2012 numbers to exceed those of 2011”.
Technology has come a long way in supporting systematic and high frequency traders. Many of these firms use highly sophisticated trading algorithms to quote and bid on thousands of securities a second. Trades are short-term, executed at millionths of a second. This has necessarily forced the exchanges to evolve accordingly as traders engage in a race to zero, driving latencies down ever further in a to exploit price inefficiencies faster than their competitors.
The London Stock Exchange bought Sri Lankan firm MilleniumIT for GBP18million in 2009 to develop its proprietary trading technology.
Nicolas Bertrand, Head of Equity & Derivatives Markets, London Stock Exchange Group, says that latency speeds vary depending on venue and geography: “Our Millennium Exchange platform consistently offers customers round trip trading speeds of around 100 microseconds. When you compare that with the 1.4 milliseconds the old TradElect platform was delivering in early 2011, you can see the significant difference in trading speeds.”
Frankfurt-based Eurex has long played a leading role as a technology provider in the derivatives space. Supporting electronic traders is hardwired into its DNA. Although it’s hard to determine the percentage of trading volume that originates from HFTs and algo traders, it is estimated that around one third of its 430 members use its co-location service.
To accommodate both the expansion in quant fund numbers, and the demands of ultra-low latency from existing clients, Eurex embarked on an ambitious project to overhaul its trading architecture. This started back in 2008 after it had acquired International Securities Exchange the previous year. It was finally implemented in 2011, and is expected to go live on the Eurex exchange this summer.
“Our technology teams are committed to delivering the best quality and superior service levels for our clients. ISE, a member of Eurex Group, won an equity options award recently and it’s using the same technology that will be rolled out at Eurex Exchange. At ISE, median latencies are in the order of 200 to 220 microseconds for the order/quote door-to-door,” comments Wolfgang Eholzer, Head of Trading System Design for Eurex.
Given the hardware demands of trading at low latency, firms sometimes need to chop and change the technology providers they work with. Da Vinci Invest is a Swiss-based HFT firm run by Hendrik Klein. Its K2 Tachyon fund trades by scanning newsfeeds for econometric indicators, including anything from nonfarm payroll data to the latest US unemployment figures in what is essentially an event-driven, albeit highly systematic, trading strategy.
Klein comments that one of the challenges the firm faced recently was the fact that its managed services provider was using too much hardware. “There were too many pre-trade risk checks which meant our latency was too high. It was taking us 15 milliseconds to execute an order in the market. It’s important that there’s as little hardware involved as possible.
“Our competitors are optimising every microsecond and we need to be at least as fast as the 10 fastest trading members on each exchange. Now we’re back to below 1 millisecond.”
Today, the firm uses RJO Brien to support it one the trading side. In addition to using Need to Know News as its primary news feed, Klein says that they are currently looking to introduce an additional news feed likes Reuters, Selerity and Bloomberg as the trading strategy evolves.
“We are starting to look at corporate news. There’s a lot of market-moving news coming out every day: profit warnings, M&A activity. That’s why we need to continuously look to improve the algorithm. We’ve now started to trade FX futures and T-notes, which we recently incorporated into the strategy,” notes Klein. Prior to this, the strategy traded six futures markets: three index futures (Dax, EURO STOXX 50 and Swiss Market Index) and three German interest rate futures: Bund, Bobl (5-year German bond) and Schatz (2-year German bond).
One of the natural outcomes of the “electronification” of trading has been the need to ramp up risk controls, both within trading firms and the exchanges themselves. Recent events like the May 2010 flash crash, and brokerage Knight Capital’s rogue algorithms, which last year caused a USD440million trading loss, have raised plenty of concerns within the industry. HFT now accounts for 40 to 50 per cent of daily trading volume in US Treasuries, according to research firm TABB Group. The inference? US Treasuries could become the next flash crash victim.
“We’ve always taken operational stability and risk management seriously from the start. We have not encountered those flash crash events that have blighted the market in recent years. Risk management on all levels, from pre-trading to trade execution through to post-trade and clearing is vital. Over the years we have implemented a number of different safeguards,” comments Eholzer.
One of those safeguards is the Stop Button, which Eurex Group offers to all clearing members to be able to halt non-clearing members. But Eholzer makes an important point in saying that people will only respect such safeguards if there’s a level playing field for all participants.
“When we design these risk tools we do so in a way that ensures no one has a disadvantage when using them. At the clearing layer, we have a real-time risk management service delivering real-time margining information to clients. Members can also set limits at the clearing level, which, if breached, will back fire into the trading engine. We have volatility interruptions for all benchmark future contracts if prices decay or rise too steeply over short time periods.
“If you drive a Porsche on the Autobahn, you’d better ensure you’ve got the right brakes!” says Eholzer. Maintaining the integrity of the markets has to be the priority.
Barney Dalton (pictured) is Chief Technology Officer at London-based Aspect Capital, one of Europe’s leading CTA firms with USD6.9billion in assets under management. Although not as aggressive as HFTs – its execution algorithms are generally quite passive – Dalton confirms that its primary safeguard is its software development lifecycle.
“We have invested heavily in this over the last year with a primary motivation of reducing software risk. This has covered all areas of software development such as: initial change specification, scheduling/prioritisation, risk assessment, design, build, automated testing and commissioning.
“Our second level of defence is the use of a ‘safety valve’ concept within our platform to ensure algorithms operate within a safe envelope. These valves operate per market as well as globally across all trading and cover areas such as trade size, price, trade slippage and orders per second. At the moment this is internal to our own software but we are also looking at using a third party FIX gateway solution to further enhance our level of protection.
“The final line of defence is our executing brokers who apply agreed risk limits to our trade flow before routing to the relevant exchanges.”
Remco Lenterman is a managing director at IMC, an Amsterdam-based HFT firm. Speaking to Hedgeweek in his capacity as head of the FIA European Principal Traders’ Association, Lenterman is quick to set the record straight and debunk a lot of the myths that surround HFTs.
The most vocal critics are some of the mid-sized buy-side institutions who see HFTs as predatory, but Lenterman says although they only represent a minority of the buy-side community, they basis of their criticism stems from a lack of knowledge:
“It’s based around: ‘Whenever I place an order in the market the market moves against me so it must be HFTs doing that’. They go on panels and all they say is ‘these guys front run my orders’. In this environment it’s politically convenient because it’s what people want to hear to justify the regulations they are pushing for.”
The front running argument is something that Lenterman has heard every since he joined Goldman Sachs in 1989. At the time it was the hedge funds or brokers that were being accused of it, now it’s the HFTs. But given that HFT firms have no client business, Lenterman doesn’t see how they could have any prior knowledge of institutional orders.
“The question is, will the price of stock that they are looking to buy or sell move against them more or less than it did 10 years ago?
“Vanguard states that the reduction in transaction costs they’ve enjoyed over the last 15 years means that the average pensioner, whose funds they manage, will have 30 per cent more in their investment account at the end of a 30-year investment cycle. If you believe HFT is a new phenomenon, which it’s not, and if you truly believe that our intention is to front run orders, then you have to believe the TCA metrics will show that. But they don’t,” says Lenterman.
The other point that Lenterman takes issue with is the idea that HFTs are willfully engaging in what is referred to as “quote stuffing”, where exchanges are flooded with quotes, only to then get withdrawn. Canada has imposed cost penalties on such activity. Lenterman says that because of market fragmentation, HFTs might be required to quote and bid in 10, 15 different markets. “If I trade in one of them, then naturally I’m going to change my quote in the other 14 markets. That’s what you call risk management.”
What people tend to overlook is that when they’re not trading, HFTs still update quotes. This helps tighten spreads, making the markets more efficient.
“I understand that there’s sloppiness with algorithms, but what I can’t understand is how ‘quote stuffing’ is meant to be used for market abuse purposes. I’m not saying it doesn’t happen, I’m just saying I don’t see the economic rationale for doing it on purpose.”
Klein adds: “Politicians claim that HFTs are responsible for this and that. The reality is we are only trading in the markets for a very short period of time, providing liquidity and helping reduce transaction costs. They’ve reduced dramatically in the last 15 years because of greater trading efficiency.”
Dalton concludes by noting that some of the Electronic Communication Networks have started imposing restrictions on the types of trading they will tolerate (e.g. Thomson Reuters Minimum Quote Life).
“In most cases these seem to have been sensible steps forward that have improved stability and reduced the risk of extreme events. Done in the right way I think some sensible restrictions to ensure that bids and offers are a genuine expression of a desire to trade is a positive step forward.”
- By Category
- News from other sites
- Special Reports
- Partner events