Trading & Risk Technology

There’s nothing revolutionary about gauging consumer opinion, except when you use social media to help. Incorporating properly distilled data from the likes of Twitter and LinkedIn offers investors access to an exponentially broader slice of the population.

Imagine going from a survey of hundreds or thousands of people to analyzing data volunteered by millions. It’s like an aggregator on steroids. And that is revolutionary.

A 2010 study in the Cornell University Library found that “collective mood states derived from large-scale Twitter feeds” predicted swings on the Dow Jones Industrial Average with 86.7 percent accuracy. Researchers used two mood tracking tools — similar to what traders could widely employ someday — to gauge the collective disposition of microbloggers.

“Including this mood information leads to higher accuracy,” computational social scientist Johan Bollen told Wired following the study. “We’re presuming on the basis of what we found, if you have some kind of super-duper algorithm and you add our time series, its accuracy will go up.”

And go up it has — at least once. A hedge fund with London-based Derwent Capital Markets beat the market last summer by tallying and categorizing keywords found on Twitter such as “alert,” “happy” and “vital,” according to The Atlantic Wire.

Drilling Down to Individual Stocks

Derwent’s approach reflects a technology in its infancy. It doesn’t try to figure out why the tweets are positive or negative. That makes it suitable only for trading broad market indices such as the Dow.

And tracking social networks still isn’t the best way to predict an individual company’s stock moves. That requires traditional vigilance over SEC filings, the CEO’s state of mind and impending scandals because most social media users don’t brag about their high-performing stocks on Facebook or Tweet about their latest bargain equity.

But they do jabber at length (pun intended) about consumer products via blogs and YouTube, not to mention exposure on Flickr. That opens the door to sophisticated indirect sentiment analysis that may someday predict swings in individual stock prices.

This technique may not be as helpful if you’re tracking the manufacturer of the latest classified military fighting vehicle. But it could be great if you’re watching the manufacturer of the latest smartphone or tablet computer — or the big-box retailer that sells them.

Market sentiment gleaned from examined social media is not yet a crucial variable in the trading equation, as my fellow Trading & Risk Technology blogger Neil McGovern pointed out to tech blog ReadWriteWeb last week. The high price of this nascent analytical technology may be a tough pill for many firms to swallow, but they ignore this type of input at their own detriment.

“Stocks can go up 5 percent or 10 percent [in one] day, which … can often be because of rumors in the market,” McGovern said. “This [technology] seems to make sense to people in the markets as a way to be able to tap into those rumors and help their short-term trading strategies.”

Spotting Corruption, Risk

Monitoring social networks can also spot patterns among seemingly random events, which in turn can expose fraud. Deploying a platform with the right mix of analytical technologies offers firms the ability to uncover market manipulation and respond accordingly.

“[If the technology] is too far off for trading, maybe we would want to look at social media for risk management,” Peter Van Kleef, managing director at Starnberg, Germany-based consultancy Lakeview Capital Market Services told CNBC.com last week. “If we can just find a couple of those time bombs before they go off, we can reduce our exposure to them.”

That means climbing mountains of unstructured data; and social media tracking technology is still learning how to walk. But early adopters are already breaking in their boots, likely to work out the bugs to reap the juiciest data — and create formidable buddy lists along the way.

Even if capital markets firms don’t take up social media tracking en masse, many companies have already started tracking themselves. Facebook, blogs and discussion groups are a few ways that an enterprise can manage its image online.

This image management has evolved from individuals monitoring the Web to automated reports to real-time aware businesses. And it represents a tremendous opportunity for capital markets firms to develop this technology in parallel.

So whether starting small with risk or going big with trade modeling, this is a budding technology with a lot of promise. As we have seen time and time again, failure to adapt soon enough could turn even today’s mightiest firm into the MySpace of trading tomorrow.

Last week (Jan. 25) at the Harvard Club in New York Aaron Brown, author of Red-Blooded Risk shared some of his thoughts on risk management at a high level by focusing on an aspect of risk management that does not get enough visibility: the risk introduced by systems that are meant to reduce risk. I put Aaron on the spot at the end of the presentation and asked him to sum up his advice in one statement, and his reply was to try to simplify.

Complex systems that are made more complex to address yet another risk element are more likely to cause more risk than is addressed by adding another feature.

Aaron highlighted the key points in his presentation using cars as an analogy. The old mechanical cars (think 1970s Volkswagen minibus) had steering controlled directly from the steering wheel to the tires. The brakes actually transmitted pressure from your foot to the brake pads, etc. Today’s cars are usually electronic – pressing the accelerator sends a signal to increase the flow of fuel through injectors; brakes are suggestions (with apologies to Toyota) for a Prius that might use a combination of braking processes, etc.

What is lost, Aaron said, is road feel.

In the same way, electronic markets have eliminated the trader on the floor getting feedback signals beyond just the information send through online news feeds. This feedback elimination needs to be understood. Electronic markets are speeding trading up, making it more efficient and lots of other good things, but there are downsides as well.

This is a subject that other authors have examined as well. In Outliers, Malcolm Gladwell uses the Iraq war as an example of how centralized decision systems based on complex computer based systems made problems worse instead of better when decision makers had a false sense of clarity of events on the ground. In Adapt, Tim Harford examines how complex safety systems can often be part of the problem (e.g. 3 Mile Island and the Piper Alpha disasters).

At Sybase we see the need for alternative monitoring capabilities external to the complex trading and risk systems available today. Our real-time analytics solutions can help build these monitoring systems – using Sybase ESP to keep track of activity in real-time. Sybase RAP and Sybase IQ can work in conjunction with SAP’s new HANA analytics solutions to store huge datasets and allow them to be analyzed at blindingly fast speeds to look for patterns of behavior not visible in the execution and risk systems, etc.

This is an interesting area, which I hope to continue to examine over the course of 2012.

Everyone has an idea of what “risk” is, and it isn’t a positive one for most.

There’s the president of a New York-based financial planning firm telling The Washington Post that U.S. stocks are too risky. And there’s Friday’s BusinessWeek headline pronouncing corporate bonds as less risky than before. Both reinforce the powerfully negative stigma that surrounds risk.

But capital markets professionals have a special understanding of “risk” that doesn’t jibe with others. Risk can be good if managed correctly, according to Aaron Brown’s dramatically-titled new book Red-Blooded Risk: The Secret History of Wall Street.

“Risk is something you dial up or down in order to accomplish a goal,” Brown wrote. “Risks are two-sided; you can win or you can lose.”

Brown is quick to distinguish risk from “danger,” such as an athlete suffering an injury, and from “opportunity,” such as a baseball player pitching a no-hitter.

Dial up risk when you’re behind in the game by playing more aggressively in hopes of scoring more points. But dial it down when you’re ahead because playing it safe reduces variables and improves your chances of keeping the lead.

Conventional book lovers will appreciate that Red-Blooded Risk covers the history of risk from the first rise of quants in the 1980s, and explains itself in terms of the Roman Empire, construction of Egypt’s pyramids and more. Contemporary readers will enjoy its manga illustrations (inset), and Brown’s explanations in terms of vampires and zombies.

Brown is one of those aforementioned quants, and today he is a risk manager at Greenwich, Conn.-based AQR Capital Management — and professional poker veteran. So he knows more than a thing or two about risk in many of its forms.

“It really annoys me, the number of people who talk to their kids about risk, but never actually go to places where people are taking risk or try it themselves,” Brown told trading blog MartinKronicle. “We [Brown and his band of original quants] sought wisdom from actual risk-takers,” he said in the book, “which took us to some disreputable places.”

Wall Street was one of the places they went, honing Red-Blooded Risk’s concepts from 1987 to 1992.

Brown will return to Manhattan to be the keynote speaker at the Sybase-sponsored lunch seminar “Risk Appetite and Road Feel” at the Harvard Club of New York City in midtown on Wednesday. He will explore the effects of technology’s trading-space invasion; how we lose the feel for trading the way power steering costs us a feel for the road; and how financial firms can put that feel back into risk appetite decisions.

Following Brown, Sybase will conclude with a brief discussion outlining how Complex Event Processing technology can provide continuous analysis of rapidly changing data to provide effective strategies for monitoring and mitigating risk. Click here to register.

“Risks always refer to human interactions,” Brown wrote. “Their level must be under our control.”

This is an opportunity to view risk in a new light, beyond its pessimistic connotations. Attendees will also get a free copy of Red-Blooded Risk.

“Red-blooded refers to people who are excited by challenges, but not to the point of being blinded to dangers and opportunities,” Brown wrote. “Red-blooded people feel anger and fear and greed like anyone else, but understand successful risk taking is a matter of calculation, not instinct.”

Debate across Europe over debt crises, austerity measures, bail-out funds and credit ratings make it easy to forget that the region’s financial institutions are on the same page about stress testing and regulation, as well as data growth and latency. Results of a recent Sybase survey released Tuesday serve as a reminder.

Rapid growth of data is a challenge to risk and compliance management, said nine out of ten capital markets executives polled at the Sybase FSI Summit in Frankfurt last month. And stress testing has not addressed all of the important risks to the banking system, according to 82 percent of them.

Some results were similar to a survey carried out at October’s Sybase Financial Services Executive Summit in London. For instance 90 percent of German respondents predicted that Basel III regulations will have a “moderate” to “significant” impact on bank profitability, down slightly from a combined 98 percent of those surveyed in the U.K. And more than half of those polled in both countries expected data latency issues to consume the most significant portion of company resources.

“There is a significant amount of uncertainty and concern among participants in the capital markets community over the current state and future outlook for the industry,” Sybase’s Dusseldorf-based managing director Theo Ruland said in a statement. “These results don’t come as a surprise to us based on the continual feedback we elicit from our customers.”

Despite relative agreement on the inadequacy of stress testing, one difference in the results centered on these annual examinations of the E.U.’s largest lenders by the European Banking Authority. While 32 percent of respondents in Germany indicated that stress testing should occur at least twice a year, 84 percent of those surveyed in the U.K. said the assessments should take place semi-annually.

The first Sybase FSI Summit in Frankfurt provided an environment in which capital markets leaders could get up to speed on the latest technologies, strategies and trends in Germany’s financial markets. Sybase hosted the event and conducted the study.

The ‘trading’ (read gambling in my view ) website InTrade allows everybody and their dog to trade on future events such as who will win the U.S. Republican nomination, who will win best picture at the Oscars, and, interestingly, will S&P rate Greece in default before midnight ET 31 Dec 2012 (see: http://www.intrade.com/v4/markets/contract/?contractId=754058). Looks like the probability, based on the latest trading, is over 80% likelihood that Greece will be in default on Dec 31, 2012.

While this is a real problem for Greece (everybody should take note – the bond market runs the world!), the impact will hit some banks more than others. Stress testing this scenario (and for Spain, Italy – even France?) should be a top priority for banks around the world. Running large ‘what-if’ scenarios is expensive and complex, and this will probably not be the only time stress testing scenarios need to be run, so investing in a historical database with high performance analytics (such as Sybase RAP) will be a good long term investment – and might also help mitigate some short term pain.

A survey conduced by LowLatency.com (see results at: http://low-latency.com/blog/polls-past-and-present-just-two-clicks-have-your-say/?utm_source=house&utm_medium=email&utm_campaign=11_12-01-12) show that only a small percentage of traders and trading strategies require nanosecond latency. In the CEP software world we understand that to shave nansoconds off users are resorting to hardware acceleration – but don’t count CEP out yet as 73% of respondents were in the microsecond (41%) or millisecond (32%) range!

Emerging markets and technologies could have a profound impact on investments in 2012, a topic Inside Market Data explored on Monday. The weekly market data publication talked to Sybase CTO Irfan Khan about how the growing importance of Big Data is changing analytics.

“Companies are at a crossroads,” Khan said. “While some are looking for a quick fix leveraging the traditional Enterprise Data Warehouse (EDW) architecture, forward-looking enterprises are going beyond the conventional infrastructure.”

Some firms are investigating more federated architectures, while others examine new deployment options.

“Expanding the conventional infrastructure towards in-memory and columnar architectures will also enable enterprises to analyze both structured and unstructured data in a single consolidated environment; permit processing with a real-time characteristic; and short-wire the latency to actionable events,” Khan said.

IMD’s Max Bowie also asked Khan about new initiatives that will provide value this year.

“First, implementing a more consolidated platform that enables enterprises to perform the analysis of both structured and unstructured data in real time will be a game changer,” Khan said. “Second, in-memory computing is reaching a maturity level where customers need to make fewer trade-offs in terms of consistency and transactional semantics, when combined with a more native integration to a columnar analytics server.”

Khan also discussed investment in mobile middleware, such as Sybase’s SUP layer, as well as incorporation of new programming models, such as Map & Reduce. Click here to read the full article, “12 for ‘12: Key Themes for the Year Ahead.”

Chalk up a victory for the financial services industry last weekend in its resistance to higher capital requirements under Basel III. Institutions would still have to maintain significant buffers to prevent another round of government bailouts, but they could dip into these high-quality assets when times get tough enough.

“During a period of stress, banks would be expected to use their pool of liquid assets, thereby temporarily falling below the minimum requirement,” stated the Group of Governors and Heads of Supervision in Reuters. GHOS directs the Basel Committee on Banking Supervision, driver of Basel III reforms.

The committee’s concession is due in part to financial industry arguments that surcharges would impair institutions’ ability to lend. But a report late last year found that the benefits of buffers outweigh their damages.

The eventful weekend also saw all 27 nations comprising the Basel committee agree to peer review of their Basel III implementation. The U.S. may have been shirking its duties under the accord, “leading other members to question why they should abide by them,” noted The Telegraph.

If Basel III implementation wasn’t enough, U.S. authorities are also having difficulty meeting Dodd-Frank deadlines, according to a progress report by law firm Davis Polk & Wardwell. As we rang in 2012, 149 of the 200 rules thus far have seen due dates come and go without. And chalk up another 200 rules that are still awaiting their deadlines.

Many long term cost factors are not on the minds of most database application users when running a given relational database management system (RDBMS), according to a Sybase-sponsored study conducted by market intelligence provider IDC. “Calculating the True Cost of RDBMS Ownership and How Sybase ASE Stacks Up: A Guide for SAP Business Suite Users” found that users also tend to favor one brand over time without taking into account evolving requirements and expenses.

Firms should regularly assess the total cost of ownership of their major database applications, the white paper recommends. This includes hardware, software and staff time, as well as opportunity costs due to poor performance and other issues.

“RDBMS products are constantly evolving,” the study notes. “Thus Sybase and its competitors are challenged to continually improve their technology to succeed in addressing customer requirements for efficient, cost-effective and flexible database support.”

So the white paper also recommends that users keep an eye on other RDBMS products. They should examine the possibility of reducing costs by migrating their application data to databases running on other products, such as Sybase Adaptive Server Enterprise.

“Over a five-year period, as [organizations in the study] grow their databases, Sybase will account for relatively more database capacity, but because of its greater efficiency, it will account for less of the total cost,” the study states (see graph). “The impact of increasing the reliance on Sybase ASE is a reduction in average hardware costs per TB from $21,187 to $20,250 over five years, saving each company an average of $139,000.”

Methodology for the study included identifying, screening and qualifying database application user organizations. IDC used these firms’ experiences to model the costs of purchasing, deploying and managing various RDBMS platforms over five years.

Click here for more information or to read the white paper.

Automated trader conducted their annual reader survey, and published the results of their findings. Collated by Bob Giffords, there were some key take-aways that I found interesting. Firstly, we have to remember that subscribing to Automated Trader is quite expensive, and so only people with an interest in automated trading are likely to participate in the survey. Perhaps a self selecting sample.

Anyway, the survey highlighted an increase in end-to-end automation, with (for me) a surprising number of buy side organizations (35%) that tied 100% of machine generated signals into orders with no human supervision (only 10-15% of sell side organizations are equally robotic. It would be tempting to point to HFT as the cause of this, however more than 60% of the buy side respondents were executing at an average rate of less than second (over 45% were at less than one order per minute). The buy-side were (not unexpectedly) also less interested in being the fastest in the market – with less than 40% indicating that latency was important to their trading. The similar figure for the sell side was about 60%.

I’ve really only scratched the surface of this results from this report – I strongly recommend perusing it if you have access.