Still, hedge funds have been hit hard by losses and client redemptions as confirmed in the latest Eurekahedge Report tracking 5,593 live funds. "The markets have been incredibly volatile and unpredictable," says Alexander Mearns, CEO of Eurekahedge. "Hence, investors are craving cash."
Not everyone is suffering, however. "Although we do systematic trading, we deal only in exchange-traded futures, options and currency, where there's been no real loss of liquidity," says Franz Hartlieb, chief investment officer for Hasenbichler Asset Management, a pioneer in alternative investments in Europe. "We don't trade credit and have no exposure to single equities, so we haven't really been affected by the crisis in terms of trading."
Consequently, Hasenbichler sees no need to change trading strategies or technology.
Yet scenes of chaos abound and quantitative trading has clearly played a role. "Half to two-thirds of US equities volumes is what we would characterize as high frequency trading, consisting of statistical arbitrage and electronic liquidity providers," says Joe Gawronski, president and COO for Rosenblatt Securities. "While the fundamentals are obviously driving the overall direction of the markets and its component stocks, the wild intraday swings suggest a market that has largely decoupled from fundamentals at any given moment and is driven by short-term sentiment. There's a lot of uncertainty and fear out there." Gawronski believes that by understanding the market and monitoring a wider range of signals, like options or credit default swap (CDS) activity, for example, trader skill can still make a big difference.
Unfortunately there is more data than ever to analyze. "Huge volume spikes for equity options have completely changed the game," says Kevin McPartland, senior analyst with the Tabb Group. "September saw peaks of 30 million contracts traded in a single day. The Options Price Reporting Authority (OPRA) is asking firms to gear up to handle 2.1 million messages per second by January 2009, which equates to an average of 10 billion total messages daily. Firms need not only to capture the data, but do the necessary pricing and analytics in continually lower latencies as well."
"Part of the growth in equities is due to derivatives since traders often hedge with cash," says Gawronski. "Probably 20 percent or more of equities volume is linked to an options play, which can add a lot of color and information if understood."
This thirst for knowledge is apparent everywhere. According to Ian Domowitz, managing director for networking and analytical and research products at agency broker Investment Technology Group (ITG), traders are doing more analysis before each trade. "We have seen demand on our pre-trade analytics service rise to 100 million queries per month from 18 countries over the past year. Pre-trade is no longer just a start of day activity," he says. "People use it continuously throughout the day."
Enriching the Data
Jeff Wootton, vice president product strategy at Aleri, a complex event processing (CEP) technology vendor, sees a new generation of real-time market analytics emerging. "Firms are starting to measure fill ratios and queue latencies in real time and using that rather than just prices to decide where or when to route their orders," he says. "Proprietary traders are looking at risk and calculating volatility in terms of standard deviations and real-time profit-and-loss or portfolio valuations to determine what and when to hedge. Real-time market insight is growing hugely."
"We're working increasingly with the signal stream derived from the underlying market data, news flows and research tools," says Stéphane Leroy, head of global sales and marketing at QuantHouse, a trading solutions provider.
Robotraders are becoming ever faster, ever smarter. "Even derivatives are becoming very expensive to use outright," says Lakeview's van Kleef. "So people are doing derivative spreads to cap their risk and offset their costs, but that requires even more sophisticated analytics."
Knowledge transfer is also accelerating. "As electronic trading grows, contracts per trade fall, and alpha margins get thinner. So, traders want more information, just as in equities," says Robert Almgren, co-founder of Quantitative Brokers. "The big value added today is a quantitative analysis of the markets." Because the technology is off-the-shelf with tools for CEP, algo trading, FIX and so on, Almgren argues, "We can focus on managing knowledge."
Multi-Strategy and Adaptive Solutions
This new breadth and depth of real-time information is needed for the rising generation of complex trading strategies. "Quants now define a range of strategies calibrated to different market conditions through extensive back testing," says Leroy at QuantHouse. "Then they build algos to recognize those conditions in the markets and run the strategies in parallel or switch them in real time. These multi-strategy solutions are still quite leading-edge."
One firm to adopt this approach is Pioneer Investments, which combines multiple models for each portfolio in a "risk budgeting" environment. "Each strategy is given a budget of risk in the phase of design and period review of the product," says Diego Franzin, head of global quantitative research and management at Pioneer. Each portfolio is composed of a core portion that grants the exposure to the asset class and several alpha strategies mounted on top of if it. Pioneer has between 10 and 15 independent alpha strategies in each portfolio. "The weight of each active strategy depends on the confidence in the model, the volatility and the correlation of the strategy with the other components."
Franzin uses models with quarterly, monthly and daily frequency that are chosen to optimize macro, leading, sentiment and technical indicators. The portfolios are invested mainly in exchange-traded instruments, where there has been ample liquidity. "We include some over-the-counter (OTC) CDS instruments, but we're mostly on-exchange," says Franzin. "By channeling multiple strategies into the models through the risk budgets, it means we leverage all of our firm's knowledge."
"Quantitative strategies can be often over-designed and become parametric with the use of a large number of input variables. These strategies are consequently 'fitted' and model decay sets in quickly," says Aref Karim, CEO of Quality Capital Management. This leads to frequent re-calibration of model parameters to adjust for changes in market fundamentals that have not been tested.
"Traders are now having to change their models much quicker because the markets are so unstable," Gawronski says.
While multi-strategy models offer one solution, Quality Capital takes a different approach, using robust, adaptive models. "Our models rely on fewer parameters, not more and accept a wider margin of error," says Karim. "They are more resilient and sustainable over time. We don't look at fundamentals and our models are agnostic to markets and/or asset classes and even to specific time horizons. Instead, we model strategies based on market phenomena, not trying to identify any unique market characteristics. We work heavily with probabilities."
Automated everything
Speed, complexity and uncertainty now dominate electronic trading. "To cope, people are trying new approaches like sampling data to throttle it, switching to low-latency messaging middleware, utilizing complex event processing and experimenting with hardware acceleration," says McPartland at Tabb Group.
Meanwhile, quants are asking for full depth-of-book to get a better picture of market liquidity. Also, some are implementing direct feeds alongside consolidated feeds like OPRA to find trading opportunities in the millisecond latencies between the two. "Despite tight budgets the buy side is still spending right across the options trading value chain," says McPartland.
Once again, technology matters. "Clients now are looking more for grid computing solutions, greater parallelism, and CEP, rather than specific functionality," says Leroy. "It takes a lot of compute power to recognize events in their market context, using real-time Monte Carlo, for example. Beyond that, they want more technology and the knowledge to use it."
"People are talking about spikes that are dozens of times the normal trading levels," adds Wootton at Aleri. To cope with such peaks, he says people are turning to CEP tools, which are fast enough to keep up with the data and flexible enough to change the rules on the fly to do real-time scenarios to increase confidence.
Smart strategies are important, too, without the low-latency pyrotechnics. For Pioneer's risk budgeting platform, for example, the approach and the models were developed in house over eight years using SQL, Matlab, StatPro or C++ and they run on a conventional Unix multiprocessor server. "The technology is in the models more than in the hardware," says Franzin.
Nevertheless, speed is still driving many decisions. "Two eyes and a brain just can't keep up with the volumes, the volatility, and the complexity of change," says Leroy.
Bob Giffords is a banking and technology analyst and can be reached at . -->