• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Demand for Quants Surges as Trading Requires More Math and Programming Skills

Wallstyouth

Vice President
Joined
5/25/07
Messages
116
Points
28
Demand for Quants Surges as Trading Requires More Math and Programming Skills


In a quiet loft building in Atlanta, a team of six mathematicians and computer scientists are applying predictive models to capture short-term inefficiencies in liquid stocks. "This is not an atmosphere that is full of adrenalin. It's the opposite of that," says Adam Afshar, president of Hyde Park Gobal Investments, an Atlanta-based hedge fund manager, who contrasts this environment with the typical trading room where traders monitor multiple screens with short attention spans. "It's a place like a library where people are doing complex calculations and they need quiet so they could keep attention on a problem for a long time."

In addition to designing genetic algorithms, Hyde Park Global is developing predictive models -- models that learn from previous trades. "These are similar to models that are used to predict the movement of rockets or the movement of a cell in a laboratory," explains Afshar, who spent 12 years at Bear Stearns before starting the hedge fund company in 2003. "These problems take days, weeks and months to solve. It takes collaboration of different people and cross-fertilization of different minds to come up with these ideas."

The quant phenomenon -- and, with it, the demand for quantitative skills -- is growing on Wall Street, as the shift to program trading and algorithmic trading requires more math, technology and programming skills. "I see the role of quantitative analysis as only getting larger, and I think we're probably seeing a change in the demographic of Wall Street to reflect more technical trading," says Peter Fraenkel, director of quantitative services at Pragma Financial Systems, a developer of algorithmic trading systems in New York.

Even pure traders are more likely to be familiar with technology and know statistical techniques, adds Fraenkel, a former managing director at Morgan Stanley who sat atop an IT hierarchy that ran an analytics library for equity derivatives. "It's not that there's a new breed of quants, but that everyone is becoming more quantitative," he says.

Charles Reyl applies his quantitative skills to the energy brokerage market. Reyl, CEO of Parity Energy, was a VP in the global modeling and analytics group at Credit Suisse First Boston. While the typical quant jobs are on trading desks in banks, he says, the career potential of quants depends on their personality. "If they have a personality that is more business-oriented, then I think the world is [their] oyster," Reyl relates. "You can succeed in being very systematic and understanding certain problems that require quantitative understanding. ... These generalists would be successful any which way they want."


Quants Move Beyond Model Building

Traditional quants still are involved in model creation, but they're also moving into quantitative trading, according to Joe Long, managing director in charge of front-office technology and quantitative recruitment at INET Technologies in New York. "A lot of the quantitative hedge funds are hiring Ph.D.s to trade," he says.

In general, the role of the quant on Wall Street is growing in importance, say industry sources. "It's been gradually increasing for some time now," says Peter Cotton, CEO of Julius Finance, a private research firm and development shop in New York, who previously worked at Morgan Stanley developing tools for pricing synthetic credit derivatives. Fortunately, "There's probably more supply of quants than there used to be," he adds.

The supply of quants is greater now than in the past because more masters in financial engineering (MFE) programs have proliferated at top universities such as the University of Chicago, M.I.T., Carnegie Mellon and the University of California at Berkeley, according to Lee Maclin, director of research at Pragma Financial Systems. Maclin teaches statistical trading and statistical arbitrage and algorithmic trading at the New York University Courant Institute of Mathematics, one of the top MFE programs in the Northeast. "There are courses springing up that deal with algorithmic trading, that review the background math and market microstructure math," he relates, noting that some of the top investment banks post advertisements for jobs in the hallways of these programs. "You see that their first priority is algorithmic trading," Maclin adds.

In fact, fueled by the shrinking order sizes and profit margins in U.S. equities, algorithmic trading is one of the hottest areas in which quants are working. "You can't afford to pay a person to put in every share because the trade size has gotten so small that even a moderate order needs hundreds of fills," explains Robert Almgren, head of quantitative strategies for equities at Banc of America Securities. Almgren, who has a Ph.D. in applied and computational mathematics from Princeton University, is a researcher and professor in mathematics and computer science who joined BAS in 2005 as a tenured associate professor in mathematics and computer science at the University of Toronto. He is in charge of algorithmic trading development at BAS.

People with quantitative skills are needed to develop new models, back-test the models and program the algorithms, according to Almgren. "A lot of that is dealing with technology -- you have to be able to implement your ideas and get the data out of the models," he says.

"Everything is data driven," Almgren continues. "You have to be able to work the databases and program the trading system. And ... you have to be smart enough to get the system to do the right thing from your quantitative analysis."

And the proliferation and increased frequency of data on The Street is further fueling the need for quants in algorithmic trading, notes John Comerford, EVP and global head of trading research at Instinet. For example, "Level 2 [i.e., Nasdaq depth of book] data is about 30 gigabytes a day," he says. "We're dealing with data that's closer to what they deal with in the biosciences and the genomes and not what people deal with in standard relational database technology."


Quantitative Trading Goes Multi-Asset

As algorithms expand from U.S. equities into other instruments, the same mathematics are being applied to trading in different asset classes, says Pragma's Fraenkel. "FX is where the most progress has been made, but people are applying these optimal execution techniques to fixed income and derivatives as well," he says.

"When we first started hiring quants in 2003 in FX, people asked, 'What does astrophysics have to do with FX?'" recalls Thomas Plaut, CEO of FX Solutions, an online broker in foreign exchange that provides a global trading system for the most-liquid currency pairs. Today, FX Solutions runs a value-at-risk (VAR) model that was programmed by a Harvard astrophysicist who has skills in high-end mathematics and in building the VAR models for the firm's automated market-making system, Plaut notes. Even retail traders that hit FX Solutions' prices are building their own models and are involved in high-frequency trading, he adds.

Meanwhile, hedge funds such as Hyde Park Global create market-neutral strategies with long and short positions that also leverage high-frequency trading -- executing thousands of trades per day -- to extract minute inefficiencies from the financial markets. In Hyde Park Global's case, predictive models, generated by quants, continuously make the decisions that drive the high-frequency trading.

But while it takes a quant to identify where the gold can be found -- that is, opportunities or mispricings in the market -- the challenge is "actually getting to the site and excavating the gold," says Hyde Park Gobal's Afshar, who notes that this primarily is a technological problem. One of the big issues is telecommunications -- how to get data delivered fast enough from the exchanges or ECNs to run calculations and send orders back to the ECNs in milliseconds, he explains.

Of course, the speed and efficiency of markets in turn is adding momentum to the quantitative trading strategies trend. The traditional approach of relying on sell-side fundamental research, which everyone is following, has lost its edge, according to Afshar. "With us, the future of finance is about engineering and risk control," he says. "We think the idea that I'm a great stock picker is not compelling. It's about, 'How am I going to structure a portfolio to be efficient, to be robust, to be able to withstand idiosyncratic risk and market risk?'"

The credit market meltdown in July and August, however, caused a number of hedge funds with quantitative strategies based on complex mathematical models to sell off the same positions and report deep losses, perhaps tarnishing the reputation of quants somewhat. But, "I think there's a distinction to be made between the quantitative funds that are using quantitative models to try and predict the market and quants who work in sell-side firms primarily doing risk management," points out Julius Finance's Cotton.

"On the one hand, the quant funds may have lost a bit of varnish," Cotton says. "But the sell-side quant roles seem a lot more important now." In fact, he continues, the credit crisis could boost demand for quants as the deep losses in subprime mortgages and quant trading strategies lead firms to revise their risk models and recalculate correlations among different asset classes.


Recruiting Quants


The credit crisis aside, the growing demand for quants is not new. "Growth of quants in capital markets has been pretty much a secular uptrend," observes Paul Alapat, managing director and head of quantitative services at Amba Research, an investment research outsourcing firm in Bangalore that provides investment and analysis support services to capital markets firms on both the buy and sell sides. Alapat says the trend is five to eight years old and is driven by improvements in computing power, more reliable databases and the availability of more-complete data sets, for instance, in emerging markets.

Quants are coming into the capital markets or investment industry from academia, including faculty and fresh graduates who are qualified up to the doctoral level, according to Alapat, a former chief economist at Nomura and regional financial economist at Lehman Brothers. Investment banks generally look to hire candidates with Ph.D.s in finance, he says, because they have a quantitative background.

Chemical engineering is another popular area from which to recruit because chemical engineers do a fair amount of optimization work in their curricula, Alapat adds. "That said, this is a scarce field," he says. "You wouldn't see the majority of chemical engineers looking to jump into financial engineering."

In fact, the supply of quants is scarce, experts agree. Overseas students from Asia and Eastern Europe, Alapat notes, represent a disproportionate share of the supply of quantitative talent.

"People are starting to go offshore to find talent to India, China and Russia," FX Solutions' Plaut says. "A lot of that talent is U.S.-educated; they come here for degrees."

Those who come to the U.S. for the education often hear of others getting jobs on Wall Street or know alumni who have built successful careers on Wall Street, oberves Amba Research's Alapat. "Definitely, if you go to M.I.T., there are successful traders or research analysts" that are alumni, he says. "And then you hear of the compensation -- lots of people during the course of their studies reorient some of their ambitions."

Quants out of school just two or three years can make anywhere from $200,000 to more than $500,000 a year on Wall Street, "depending on their experience and how smart they are," notes INET Technologies' Long.

So, when quants get to The Street, on what sorts of projects are they working? Broadly speaking, quantitative analysts are working on "alpha-generation strategies that involve a lot of testing and validation work until they are comfortable that a certain strategy will produce returns over a certain benchmark," says Amba Research's Alapat. The second area is managing, measuring and monitoring risk, he adds. "If you take these two groups, you will cover the work that is done," Alapat asserts.

According to Hyde Park Global's Afshar, his firm mainly is interested in hiring computer scientists with strong mathematics backgrounds or mathematicians with strong computer science backgrounds. "Of course it helps to have somebody like me that has Wall Street experience so you are not completely in some sort of illusionary world," he says.

On the other hand, Afshar continues, it can be difficult to attract the top quants to finance. "We find it very difficult to hire the very best academics to join our field, to seduce them with Wall Street," says Afshar. "It's because money is not the primary motivating factor." Afshar says he finds that quants need to be "seduced by the technology, by the field, by the questions you suggest to them, by the idea that they are working at the epicenter."

So is the quant the top job on Wall Street? "It's the top entry job on Wall Street," according to FX Solutions' Plaut. "The top hedge fund advisers surround themselves with people who have these quantitative skills," he says. "If you look at Citadel and Renaissance [Technologies], both of those firms were started by people with significant quantitative skills and they've been extremely successful."

Instinet's Comerford says he looks for people who can span multiple dimensions. "Almost everyone has a strength in quantitative analysis, trading or technology, but everyone needs to be very strong in at least two" of these areas, he says.

Going forward, the demand for quants is bound to increase as the trend toward high-frequency trading and cross-asset class strategies creates the need for new models and algorithms. "On both sides of the aisle, the buy and sell sides, quants are becoming more important," FX Solutions' Plaut comments. "The cult of personality is really being replaced with the cult of the scientist -- the quiet person sitting behind his desk working on algorithms and trying to find efficiencies in the marketplace and efficiencies in the back-office processing of trades."

By Ivy Schmerken
 
thanks for a good read.
what maes me curious is that the emphasis is made on PhDs and pure science background (ee, chem eng, etc), they only briefly mention mfes. i wonder if we fall somewhere in the middle. and if yes, what is that 'middle'?
 
Real good read. Very true from what I've been able to see so far specially from the quantitative groups. Everyone is using the same data, signal so it's really hard to gain an edge. Even the fundamental, deep value guys begin to dig into historical data to find some edge.
"Everything is data driven," Almgren continues. "You have to be able to work the databases and program the trading system. And ... you have to be smart enough to get the system to do the right thing from your quantitative analysis."

And the proliferation and increased frequency of data on The Street is further fueling the need for quants in algorithmic trading, notes John Comerford, EVP and global head of trading research at Instinet. For example, "Level 2 [i.e., Nasdaq depth of book] data is about 30 gigabytes a day," he says. "We're dealing with data that's closer to what they deal with in the biosciences and the genomes and not what people deal with in standard relational database technology."
 
thanks for a good read.
what maes me curious is that the emphasis is made on PhDs and pure science background (ee, chem eng, etc), they only briefly mention mfes. i wonder if we fall somewhere in the middle. and if yes, what is that 'middle'?

good question Dmytro, can some one attend to it.
 
I wonder if it is possilble to offer a course on Statistical Arbitrage at Baruch :)
 
No MFE covers StatArb anything like well enough, so there is definitely a gap in the market.
Also as a HH, I perceive demand that seems to have increased monotonically throughout turbulent times.
In StarArb/Algotrading a quant is closer to the money, which is also good.

Highest paid entry level person we ever dealt with was a SS/AT, during the bidding, we realised that we were going to have fun when GS folded out.

But more than any other type of role, you need to be good at C++, Baruch is ahead of the mean by doing this, but you need more than this to even get me to put you forward.

Signal processing is good, if I were Prof S. I'd get the EE guys to do this for you, indeed, with all due respect to your lecturers, I would insist on experience from that discipline.

You also need econometrics & time series of course.


But in return for the higher mean, you are pushing up the variance, and the hurt is in the tails.

SA isn't great for helping you get into risk management and you are closing the doors to a lot of safe options.
 
Ideally, I would like to see Econometrics, Time Series (by MFE program), Statistical Arbitrage, Algorithmic Trading offerd at Baruch.

Dominic, could you please explain a bit more about Signal Processing. What part does it play in finance, how important it is?
 
Financial time series are so noisy that classical finance theory started with assuming that it was entirely random.
However there are models that come from market microstructure where you can detect processes going on and exploit them.
Also there are signals in terms of movements in underlying or correlated assets which imply direction over a short term.

Even a simple single factor mean reverting interest rate model requires you to extract "signals", ie the mean and reverting factor. Also, of course interest rates jump to being around other means, and this typically correlates with the factor changing.

Spotting regime changes of this form is pretty tricky, and signal processing offers some approaches.

Even the volatility that underpins pretty much everything we do is heartbreakingly hard to determine. Black Scholes has it as a continuous function, but we have discrete noisy samples.
Take "current" vol by taking different windows, and you get wildly different values, and that is before we enter the faith-based hokum of volatility forecasting.
 
Dominic, thank you for your input. I've heard about signal processing being used in finance but was not aware of where and how. Someone who I know is familiar with signal processing but not familiar with finance, maybe we will benefit from each other.
 
Back
Top