Universities fear power of robot trading to lure maths talent

Norwegian universities are fighting an uphill battle to hold onto talented mathematics, physics and technology students due to demand from the finance industry, academics have warned.

Professor Fred Espen Benth of the department of mathematics at the University of Oslo said: “We increasingly are losing good master candidates to the finance sector in Oslo. I think that the trend in the US to hire ‘rocket scientists’, which includes mathematicians and not only rocket researchers, has been adapted in Norway.”

The Norwegian media has reported that doctoral maths students at the university are being siphoned off by large banks, which increasingly rely on computers to make decisions based on huge amounts of electronic information too large for human traders to digest.

The introduction of robotic stock trading by super-quick computers based on algorithms has made mathematicians, physicians and IT-engineers the new stars of the stock exchange.

Some estimates say that up to 90% of buying and selling stocks is now carried out by ‘algo-trading’, ‘robot-trading’ or HFT – ‘high frequency trading’. The latter uses complex algorithms to execute orders based on market conditions. It requires high execution speed and robust algorithms.

Algorithm-beating traders in court

The trend was highlighted by a recent case in the Norwegian high court, in which two ‘day traders’ appealed against suspended sentences for manipulating the Oslo stock exchange.

They had beaten the algorithms for the robotic trading run by Timber Hill in the United States – for five months.

One of the traders Peder Veiby, a 27-year-old student of BI, the Norwegian Business School in Oslo, told Hegnar Online that he could have beaten the robot for NOK100,000 (US$16,700) a day for a long time. He also estimated his yearly profit now from trading at the Oslo Stock exchange to be NOK500,000 (US$83,600).

In 2007, Citigroup – now one of the major firms in ‘algo-trading’ in the world – bought the Rutgers university spinout firm Automated Trading Desk for US$700 million. The spinout only employed 100 people but still managed to trade 6% of the trading volume of major US stocks, estimated to be 200 million shares per day.

Talent being lured away from universities

The fear among Norwegian universities is that demand for this type of expertise is luring talent away from universities – and it appears to be a global phenomenon.

Yngve Ness, who worked as an institutional sales trader at Alfred Berg Inc in New York and at ABG Sundal Collier Ltd, said the need to cut costs in the current climate, combined with the need for expertise, may mean that firms around the world are increasingly turning to young academic talent in maths.

He said: “Since the mid-2000s maths-based automated trading systems have developed and now dominate large parts of global equity trading. Institutional clients, spanning pension and hedge funds, have increasingly turned to various ‘machine’ facilities.

"Stockbrokers and independent vendors have developed electronic crossing networks that provide ‘dark pools’ of liquidity that eliminate market impact. More commonplace are the algorithm-based platforms suitable for a variety of forms of order execution.

“Cost savings and anonymity are clearly key attractions for the client,” Ness said.

Traditional stockbroking services are still in demand, as fund managers still want expert sales, sales trading and research input from their brokers. But Ness said the difference was in the payment.

“In previous years brokers would receive commission based on executed order flow. Order flow is the lifeblood of any trading desk as it represents the basis for traditional broking; matching and crossing client business. Now there are robots, fewer real tickets in the hands of real brokers, more cheques in the post and a lot more maths involved,” he explained.

Peter Warren, the Norwegian hedge fund manager of Warren Capital AS in Oslo, said one HFT company he knows in London spends on average US$45,000 a day, 250 days a year, to secure the fastest processors. In addition there are software costs.

“Robots-HFT trading is [following] ‘big boy’s rules’: if you do not have the capital, you have to find another way to trade,” he wrote on his blog. “The British investment management firm Winton Capital Management has 200 physicists and mathematicians among their staff.

“That tells us something.”

Problems in other European countries

The brain drain to the finance industry is being felt across Norway’s border with Sweden.

Professor Uno Fors, of the department of computer and system sciences at Stockholm University, said it had a problem with many students exiting bachelor and masters studies before they take the final exam.

“This is due to them getting well paid work in the industry without having a degree. And it is true that many of the best students disappear first. So we have a similar problem. But we do not know where they go. Some might end up in major banks or finance, but we do not have precise data.”

In Germany Professor Peter E Kloeden of the Institute of Mathematics at Johann Wolfgang Goethe-University of Frankfurt am Main, co-author of a renowned book on numerical methods for stochastic differential equations and author of numerous articles on issues of relevance for financial mathematics, said most of his masters students went to banks rather than doing a PhD.

But he said there were payoffs from the draw of jobs in the insurance and banking sector too, as many students want to do their bachelor and masters projects in finance or computational finance rather than in some more mainstream area of mathematics; and a lot of students do praktikum at banks or finance firms during their studies.

Similarly, Oxford University in the UK has seen the demand from finance as an opportunity rather than a threat.

Professor Brian Stewart, in the five years before he retired in 2009, was associated with the Skills Agenda programme in the university's mathematical and physical sciences division. It used to put on workshops for postdoctoral and postgraduate students on building a career, either in academe or in the wider world.

“At that time the financial services sector was certainly keen to employ at this level from these areas of study. From the sector we always had representatives from the investment managers, looking for people with very high-level mathematical skills,” he said.

The university has responded by creating new courses aimed both at current practitioners who for compliance purposes need a technical qualification, and those wanting to make the transition.

Daniel J Guhr, managing director of Illuminate Consulting Group, said the demand for maths and physics majors had been around for some time. He said his peers at Oxford were hired by banks, investment firms and hedge funds to build quantitative models back in the 1990s.

Wall Street has done so for a long time as well. Google has become the biggest recruiter of high-quality IT engineers, hiring 3,000 or more a year.

A good number of these graduates would have gone on to graduate school or PhD programmes, and some would likely have become faculty members.

Some students actually run their own trading operations on the side, or try to set up shops doing so. Guhr said there was a simple reason for this: the choice of leaving academia to earn “US$150,000 or much more” is just too tempting.

Others such as Maurits van Rooijen, rector magnificus at Nyenrode University in Amsterdam, have argued that the availability of highly attractive job options will make it easier for maths departments to attract gifted students who might otherwise think there would be limited choices other than a career in teaching.

Window of opportunity closing?

But according to Peter Warren, the HFT window for mathematicians is closing, just as it did when the electronic trading systems for foreign exchange opened for direct algorithmic interfaces in the late 1990s, as only the biggest companies would be able to justify the expense of continuously speeding up their processors.

Another element threatening high frequency trading is the fact that stock exchanges are beginning to penalise many such activities, raising their cost of trading substantially and thus cutting profitability.

He believes the market for mathematicians in HFT will become saturated and mathematicians straight out of school will not be able to identify new opportunities, so their services will be needed only after experienced market participants are able to identify them.

But Oslo’s Fred Espen Benth does not think that robot trading is the only factor behind the increased demand for graduate candidates in maths, physics and IT-technology. It could be that the finance industry is more quantitatively complex, and that it sees the need for people who are analytically trained.

“We presently have one PhD position vacant, where I have three very good students that I have asked to apply, but only one did so. The others are going to deliver their master thesis before the summer, and they do not have a job offer yet.

“But since they have not applied, I interpret this as meaning that they are targeting better paid positions out there that are more attractive than a PhD with us,” said Benth.

“That is sad.”