# Article: High-Frequency Trading Firms Seeking Tech Talent

#### Andy Nguyen

One of the most closely guarded secrets in the competitive world of high-frequency trading is how these black-box firms go about recruiting talent. With high-frequency trading dominating the markets, there is increasing demand at HFT firms for workers with the right skills, especially programmers, strategy builders and quants who can work with complex data sets and build high-performance trading systems. Some observers even believe that the real battle lines on Wall Street will be drawn over hiring programmers and developers, as opposed to traders.

"There's a big push in the space to find people with physics, heavy computer science and quantitative finance degrees," relates John Netto, president of M3 Capital, a proprietary trading firm in New York that also serves as a consultant to the alternative investment space. HFT firms want "big, strong mathematical backgrounds to put together these strategies."

According to Petter Kolm, deputy director of the mathematics and finance program at New York University's Courant Institute, whose master of science of mathematics in finance program specializes in quantitative finance, "What they need is people with very strong programming skills, so they get them directly out of schools or through other channels. It's called 'real-time system experience.' These are people who have the experience of writing code and analyzing huge amounts of data and the ability to move data back and forth quickly. Those types of generic computer science skills are always attractive."

But with the proliferation of proprietary trading shops and hedge funds that leverage high-frequency strategies as well as agency brokers doing low-latency algorithmic trading, such candidates are difficult to find, experts say. Joe Long, a quantitative-trading executive recruiter at I-NET Technologies in New York, says a programmer who is good at making a system run very fast in C++ could make as much as $500,000 at a hedge fund. "We do a lot in the$300,000 to $600,000 range with a developer that understands Unix. The technology skills are difficult to find," he reveals. "Good programmers are hard to find," confirms an executive with a proprietary trading firm in Chicago who spoke on the condition of anonymity because he did not want to reveal any proprietary information. "That's why people spend so much time and resources finding them. They command big salaries, and they are highly competitive." And it isn't only the large ultralow-latency firms that are seeking these skills. "We share the same technological burdens and data flow requirements [that] the ultrahigh-frequency shops do," says the executive, who notes that while speed is important to his firm, which specializes in options and options volatility trading strategies, it is not in the same camp as "ultra" high-frequency shops such as Getco or Sun Trading. "We're also dealing with the massive data crunch, processing markets, and analytics and strategies." From Ivy Halls to Wall Street To find qualified candidates, the firm often recruits from master in financial engineering (M.F.E.) programs around the country, the executive reports. The firm also targets candidates with straight engineering backgrounds. One of the benefits of M.F.E. program graduates is that they "have both the higher level math and engineering skills so they know how to build things," he says. Many HFT firms search for the best and brightest by recruiting out of top universities, such as M.I.T. and Carnegie Mellon. "You go to the big technical engineering schools -- it's not any secret that people are trying to tap that," the executive says. In some cases, he notes, high-frequency trading players, along with proprietary trading shops and trading software firms, even are donors to the engineering and quant finance schools, helping them assemble trading rooms. The benefit of hiring young minds is that they look at problems in different ways, sources say. But if firms hire fresh talent from the top M.F.E. programs, they will need to train them in the details of the business. If they're designing systems for options trading, for example, "You do need to know what puts and calls, deltas and gammas are," the Chicago prop trading executive says. In his firm's case, "[New hires] are mentored by senior people in the firm," he relates. "We need to get them up the curve in what types of problems exist in the business." Other firms also have internal training programs (see related sidebar). In addition to recruiting at top universities, computer-driven trading firms also hire experienced developers from other industries. On the networking side, firms might target talent from a company such as Cisco or another big telco firm, "because you're moving packets of data and switching," the executive relates. But, he cautions, network professionals coming from other industries may not be familiar with the speeds that high-frequency shops require. "No one is dealing with the kind of speeds in trading that these high frequency guys are. Even if you take someone who ran big networks at big banks, the transactional speed is not as high." Of course, high-frequency trading firms also are hyper-competitive with each other, and often they will lure talent away from a competitor. For example, several industry sources cite HFT start-up Teza Technologies, which hired Sergey Aleynikov, the ex-Goldman Sachs programmer who is accused of stealing code from the investment firm's stock and commodities trading platform. Given the demand for at sub-millisecond and even microsecond executions, one can understand why Aleynikov's skills were in demand: His bio on LinkedIn claimed he was responsible for development of a distributed real-time colocated high-frequency trading platform. On the social networking site, Aleynikov described the platform as "a very low-latency (microseconds) event-driven market data processing, strategy and order submission engine." There also is speculation that Aleynikov was of interest to Teza because he had previously worked at IDT, a telecom company in Newark, N.J., where he had experience with order routing and switching -- that experience may have been even more valuable than what he had learned about the Goldman trading system. With Aleynikov's knowledge, Teza reportedly was willing to pay him$1.4 million, more than triple his \$400,000 pay package at Goldman.

It's Who You Know

Given the risks inherent in hiring new employees (as the Teza-Goldman case suggests) and the tight-knit nature of the HFT community, however, people tend to ask colleagues if they can recommend anyone -- who maybe was let go recently -- who has the right skills, sources say. And since brokers build relationships with people in different firms over time, they are another source of recommendations in the recruiting game.

"You're looking for a very unique skill set," notes Brian Tahan, VP, Advanced execution Service (AES), Credit Suisse. "If you throw this out on Monster.com, you're not going to be hit with [qualified candidates]. It's also difficult to walk into M.I.T. or Stevens Institute of Technology and ask, 'Who do you have?' "

But it's not clear that high-frequency trading firms prefer to poach talent from Wall Street firms. Several recruiters say that some quant firms, such as Getco and Renaissance Technologies, don't hire from Wall Street.

"They'd rather hire developers from outside the financial industry -- from Google, Microsoft and Amazon," says Edward Guy, managing partner at Nations Staff, a financial technology recruiter in New York. "They like those guys who work with very fast systems and are more about pure development." Guy adds that automated trading firms want people that think outside the box.

Many HFT firms believe that it's easier to teach candidates about the business than to teach them about technology. "That's not uncommon," acknowledges Credit Suisse's Tahan. The idea is that candidates who have master degrees and Ph.D.s in scientific fields have the aptitude to pick up the finance side. "Whether that's true or not, that's the reason for the decision [to go outside the industry]."

Adds Nations Staff's Guy, "The new hires don't need to know about the financial markets. It could be that they build systems for books, airline tickets or stocks. As long as you have the right mentality."

In 2007 James Simons, CEO of Renaissance Technologies and a former math professor, publically stated that the hedge fund manager only hires scientists (e.g., mathematicians, astronomers and computer scientists) to develop its automated trading strategies. "We haven't hired out of Wall Street at all," he said in a speech to the International Association of Financial Engineers, according to a Reuters report.

Supply and Demand

Another reason for looking outside the capital markets industry is simply because that's where the available technology talent is. Three years ago trading firms would have had to compete for these candidates with mortgage desks, which were hiring 50 at a time. But with the implosion of the real estate market, there now is larger supply of willing candidates. "That helps shape the strategy too -- where the availability is going to be," the executive from the Chicago prop trading firm says.

And with high-frequency trading in the headlines this past year, many more potential candidates than before are aware of the high-frequency trading specialty -- and of the profits HFT firms are raking in. "It's got sex appeal," he says of high-frequency trading. "Whereas three years ago, the high-frequency trading firms may have had a hard sell, it's probably getting easier because of the way the market has transformed."

Still, quantitative trading firms have very high standards and can be extremely selective in whom they ultimately hire or even grant a telephone interview to. "We need people who are sure bets, and we need people who will be able to hit the ground running," says Peter Fraenkel, chief technology officer at Pragma Securities, who adds that Pragma, a pure agency brokerage firm that executes trades on behalf of its clients, is highly quantitative and develops algorithms and optimal execution services. As such, the firm cannot take advantage of narrow skill sets, which can make recruiting challenging.

"It's not enough to hire somebody who has abstract math skills or happens to do well in technical subjects," Fraenkel explains. "They also have to be adept at using computers and software tools to handle the massive amount of high-frequency data that includes every single tick and quote that occurred during years of trading days. So this isn't regular database skills. Our quants have to program 95 percent of the time, whereas a developer has to program 100 percent of the time."

Going forward, with the growth of high-frequency trading and algorithmic trading strategies overall, sources expect there to be ongoing demand for technology talent. But this doesn't mean that high-frequency shops will be go on a hiring binge.

Unlike some of the biggest players -- including D.E. Shaw, SAC Global and Renaissance Technologies, as well as the proprietary trading desks of bulge-bracket firms such as Goldman Sachs, Morgan Stanley and Deutsche Bank -- most high-frequency trading shops are smaller boutique firms, so they don't need to hire that many people. For instance, Getco, the Chicago-based electronic market maker that does 15 percent of the volume in U.S. stocks, has just 225 employees, according to the company's spokesperson.

"These small businesses tend to hire more under the radar screen," explains New York University's Kolm. "It's more like cherry picking, and they tend to reach out to their informal network through contacts and asking people for references."

In fact, while there have been a number of high-profile moves lately by Wall Street executives leaving exchanges for high-frequency trading shops, some sources say the developers are becoming more important than the traders. "The more important [hires] are the guys who you never heard of, building feed handlers and other parts of systems," says one source who spoke on the condition of anonymity.

"In our world," adds the Chicago prop trader, "There are many instances where a strong developer would be more valuable than a strong trader."

#### IlyaKEightSix

So for those of us who aren't heavy coders? We're screwed going forward? I mean I'm using R to literally do all my statistics HW nowadays, I know Excel, and have seen Unix and such, but I'm not some top-tier super-dee-duper programmer (and most people aren't, either) because they thought being a quant meant doing *quantitative* stuff not programming stuff.

#### Eugene Krel

##### sunmulA
Yes, if you refuse to adapt you are in fact screwed going forward no matter how much R you use to do your homework.

#### IlyaKEightSix

It's not about refusal to adapt so much that I never had much talent programming. I could think logically, but when trying to speak to the computer, things go south quickly.

If it was as easy as simply downloading a few e-books and trying to hack some C++, that'd be one thing--but when I was an OR major and not a comp-sci major who only realized that actuarial science was a scam and a sham in my senior year...I don't exactly have a time machine to tell myself to go back in time and DL Savitch's Absolute C++ before my first programming course and ace all my coding classes from then on (the first three of which I did horribly in).

The real question is, for the vast rest of us, what do we do to compete vs. those top-tier coders?

#### DanM

##### Math Student
The real question is, for the vast rest of us, what do we do to compete vs. those top-tier coders?

We can start our own HFT firm. If you can't beat them - hire them.

#### chaoticrambler

It's not about refusal to adapt so much that I never had much talent programming. I could think logically, but when trying to speak to the computer, things go south quickly.

If it was as easy as simply downloading a few e-books and trying to hack some C++, that'd be one thing--but when I was an OR major and not a comp-sci major who only realized that actuarial science was a scam and a sham in my senior year...I don't exactly have a time machine to tell myself to go back in time and DL Savitch's Absolute C++ before my first programming course and ace all my coding classes from then on (the first three of which I did horribly in).

The real question is, for the vast rest of us, what do we do to compete vs. those top-tier coders?

You have to shine somehow or the other...you can bedazzle the banks and get a traditional quant job doing stochastic calculus and other quanty stuff. Or you can become a master C++ programmer and work for these hedge funds.

#### doug reich

##### Some guy
You have to shine somehow or the other...you can bedazzle the banks and get a traditional quant job doing stochastic calculus and other quanty stuff. Or you can become a master C++ programmer and work for these hedge funds.

To build that sentiment, it doesn't make sense to get excited every time one corner of the business tells you one thing and the other tells you another. Are you on the CFA forums fretting about your lack of accounting knowledge?

The best course of action for you or anyone else is to educate yourself about what is in demand from the people who you are actually interested in working for. Did you want to be a high frequency software developer before being told that you can't? If you read that article, it takes a good 5+ years of software immersion to get the jobs they're talking about.

What do you want to do? Go find the people doing it and listen to them.

#### GoIllini

##### Market Crises= Gray Hair
It's not about refusal to adapt so much that I never had much talent programming. I could think logically, but when trying to speak to the computer, things go south quickly.

If it was as easy as simply downloading a few e-books and trying to hack some C++, that'd be one thing--but when I was an OR major and not a comp-sci major who only realized that actuarial science was a scam and a sham in my senior year...I don't exactly have a time machine to tell myself to go back in time and DL Savitch's Absolute C++ before my first programming course and ace all my coding classes from then on (the first three of which I did horribly in).

The real question is, for the vast rest of us, what do we do to compete vs. those top-tier coders?
My impression is that it's not that big of a deal. In fact, one of those high-speed coders spent a few days upgrading an application I work on and then threw our production environment through a loop. This had to be cleared up by one of our first-years. These guys are smart and specialized, but that expertise doesn't necessarily extend into other areas of development.

On a large scale, writing fast code is a matter of reducing the algorithmic complexity and figuring out ways to cache/memoize the expensive steps. On a small scale, it has to do with understanding how the compiler will convert everything into assembly and maybe even how the CPU will pipeline everything.

At a large firm like a major investment bank, for something where speed is really important, you can have people who specialize in both of those areas and the quant helping to put everything together; it's difficult to be an expert on assembly and algorithms at the same time.

#### chaoticrambler

To build that sentiment, it doesn't make sense to get excited every time one corner of the business tells you one thing and the other tells you another. Are you on the CFA forums fretting about your lack of accounting knowledge?

What do you want to do? Go find the people doing it and listen to them.

No, I am not on any CFA forums or anything. I am on Wilmott with same id but thats it.

I am all for not buying into all the hype and understanding my competencies and limitations.
I am engineering PhD. student with a research involving analysis, numerical algos, probability and differential equations and I am working out stochastic calculus these days, and that is my focus right now. I am not trying seriously for algo-trading positions.

#### voltronhh

No, I am not on any CFA forums or anything. I am on Wilmott with same id but thats it.

I am all for not buying into all the hype and understanding my competencies and limitations.
I am engineering PhD. student with a research involving analysis, numerical algos, probability and differential equations and I am working out stochastic calculus these days, and that is my focus right now. I am not trying seriously for algo-trading positions.

you don't have to be the most amazing coder, but you should be competent. if you're smart enough to be a quant, you can learn how to code. knowing practical things about computers (basic IT skills) is useful to. If you're not good with computers, it's hard to put your ideas into production, you'll only be able to do theoretical work. you won't be as useful in a practical sense.

#### ExSan

So for those of us who aren't heavy coders? We're screwed going forward? ...
yes, you are right
BTW worth to read AN post, thanks !

Replies
0
Views
341
Replies
35
Views
6K
Replies
0
Views
4K
Replies
12
Views
848
Replies
1
Views
2K