Photo: Avi Shauli
Trading through algorithms(illustration)
Photo: Avi Shauli
Mathematical algorithms - most important asset (illustration)
Photo: Index Open

The rise of cash machines

Calcalist Magazine celebrates first anniversary with in-depth look into Algo-Trading – one of most confidential and profitable industries in Israel

On Sunday, October 18, 2009, four of the most successful businessmen in Israel landed at Ben-Gurion Airport. They returned home from New York where they had attended a business meeting that went unnoticed by the press. No one recognized them as they proceeded to reclaim their baggage, because their photographs have never appeared in any newspaper.


Their company, which was first exposed in daily business newspaper Calcalist in June 2009, but has been sparsely mentioned since – managed to keep a low profile, and earn approximately $230 million in 2008 and in 2009 approximately $170 million. And if the four anonymous passengers had their way, no one would ever have heard of them.


Noam Stern-Perry, 38, by education an electrical engineer; attorney Yoav Ben-Dror, formerly a partner in S. Horowitz, a prominent law firm; and their partners Rami Lipman and Edi Shapiro were four of the seven major shareholders of Final, a private company which in the past two years has been among the ten most profitable companies in Israel.


The other partners are the company's former CEO, Yariv Gilat; Arie Kelkstein; and the man who keeps the lowest profile among them, Final's founder Nir Kelkstein, who, based on estimations obtained by Calcalist has earned over NIS 500 million ($140 million) from his engagement with the company. Such earnings make him one of Israel's 500 richest people.


Final is a private company which through a new and controversial method of trading has outperformed the stock exchange market since 2001. Kelkstein and his co-founder Stern-Perry own, collectively, approximately 40% of the company's shares and employ approximately 150 workers, most of whom are former stock traders, mathematicians and computer programmers.


A quick glance at those folks would not reveal any of this. Kelkstein, 38, lives a modest life compared to the wealth attributed to him in a small village 20 km outside of Tel Aviv. Like his partner Stern-Perry, he avoids publicity and any interaction with the media.


Now it transpires that Final's exposure in Calcalist was but the tip of the iceberg. Final's senior managers are not the only ones who avoid publicity. Tens of alumni of the IDF's technology units, experienced stock exchange traders, mathematicians and Hi-Tech people started, in the past years, to form companies that attempt to repeat the success of Final and of similar companies operating across the world.


Like Final, such companies keep a low profile, away from the watchful eye of their business competitors, the media and the regulators. Their offices are located in places such as a private house in Ramat Hasharon (small town north of Tel Aviv), behind a black unmarked door in the State Square in Tel Aviv, or in a luxury apartment complex by the beach.


Most of those companies will not survive, though their founders believe themselves to be pioneers in a new area of investment which will transform the trade in securities. Those few who do make it may earn more than any investor who has ever operated in Israel. Calcalist Magazine now reveals the scope of the secret industry in Israel.


'The trade of the carnivores'

On July 3, 2009, Sergey Aleynikov found himself surrounded by FBI agents in the middle of Newark Airport in New Jersey. Despite the striking similarity to old espionage movies, Aleynikov was no Russian spy but rather a programming engineer who had spent the previous two years working for Goldman Sachs.


A few months before his arrest, he had left Goldman for "Teza Technologies", a small, little-known company from Chicago, which offered to triple his paycheck from an annual salary of $400,000 to $1.2 million.


According to the charges brought against him, prior to his move to Teza Aleynikov broke into Goldman's computers and downloaded lines of a mathematical algorithm code which the bank's research department had developed.


As Goldman's representatives subsequently informed the FBI, the algorithm was a part of a program which analyzes thousands of global reports and pieces of information concerning securities in real time, based on which it conducted dozens of quick sale and purchase transactions, designed to preempt, by fractions of a second, the other investors in the market.


According to the Wall Street Journal, which has been following the affair, this sophisticated trading program enabled Goldman to conclude Q2 of 2009 with profits of approximately $3.4 billion, and Q3 with profits of approximately $3.2 billion. In Q2, the bank demonstrated unprecedented performance: in 46 days, constituting 71% of the trading days of that quarter, the bank made profits in excess of $100 million and only 2 trading days resulted in losses.


The arrest at the airport made headlines of the fast-spreading and as-yet-unpublicized domain of trading through algorithms, or, in short, "Algo-Trading".


Algo-Trading was first introduced in the 1980s and started developing in the past decade. In the past three years, it gained momentum both in Israel and abroad, while the leaders of the trend have attempted to keep as low a profile as possible.


In Algo-Trading, the stock trader is replaced with a powerful computer. The computer collects real-time information on shares, analyzes it within a fraction of a second and makes, based on such information, sale and purchase decisions and carries them out through thousands of instant instructions which are delivered to the stock market.


The analysis is conducted by mathematical algorithms, developed by Algo-Trading companies, and considered the most valuable asset in that secret industry. Delivery of instructions is done through globally-marketed programs, known as "High Frequency Trading" or "HFT".


No human trader, and barely any trading system, is capable of outpacing the computers of the Algo-Trading companies. Once information which may result in a hike of a stock price is published – for example, a change in demand in the market, a positive reference in the media, or even indirect data such as minor changes in raw materials which serve the company that issued the stock – the Algo-Trading computers are the first to identify such information, purchase it a fraction of a second before the rest of the market, raise its price and then make a more profitable sale than those who purchased only a moment later.


Through such methods, Algo-Trading companies can exploit opportunities and make maximum profits, many times at the expense of other traders. Such features induced the media to ascribe a unique nickname to Algo-Trading – "Carnivore Trading".


Aleynikov is not the only mathematician who received an attractive offer from a company in the industry. In recent years, Algo-Trading companies in the US have recruited graduates of the Mathematics, Physics and Engineering Departments of Princeton, MIT and other universities, just as NASA and the US Military did in the past. A source in a senior position in a young Israeli Algo-Trading company told Calcalist Magazine that annual salaries in the US are in the range of millions of dollars and some have reached tens of millions.


Goldman Sachs is only one of the leading powers in the Algo-Trading industry. According to Rosenblatt Securities, a NY-based consulting firm, the four leading forces in the field are Medallion, Citadel, Getco and Goldman Sachs.


Medallion is a hedge fund, owned by the mathematician James Simons, who is considered the pioneer in the field of combining mathematics and capital markets. According to estimates, Medallion's revenue from Algo-Trading in 2008 reached $2.5 billion. Citadel is a Chicago-based market maker with HFT annual earnings of $1 billion. Unlike Citadel and Goldman, which ventured into the industry after years of more traditional trading, Getco, which is also located in Chicago, is a company based on the idea of transforming mathematics into money, and managed to conclude 2008 with approximately $400 million in earnings.


Other pivotal names in this rapidly growing sector are Sun Trading, Two-Sigma and Hudson-Trading, which are located in NYC, while in Europe the most dominant companies are Dutch Optiver and Czech RSJ.


These companies stand behind the rapid growth in artificial-intelligence-based trading around the world. According to a report published by AITE, a Boston-based consulting firm, Algo-Trading companies are responsible for 73% of the trade in the US. Another report, published by TABB, a research company based in the Boston suburbs, asserts that 2008 – the year of the crises, in which investors in the NYSE lost $8.5 trillion and many hedge funds and investment banks collapsed – was great for the Algo-Traders, who recorded an aggregate profit of $21 billion.


Following such publications, the SEC announced that it will examine the legality of the field. The investigation commenced in July, and two months later the SEC announced the first limitation: a prohibition on placing Algo-Trading computers in the buildings of the stock exchange markets.


According to the SEC, the physical proximity of the Algo-Trading computers to the stock exchange computers shortens by a thousandth of a second the time that the information takes to travel from the stock exchange to those computers, and due to their excessive speed, it is considered receipt of prior information.


Algo-Trading companies are concerned that the SEC will ultimately determine that the operation of the Algo-trading computer, from any location, is akin to receiving prior information, which may result in additional and stricter restrictions.


One of the restrictions which have been considered is the cancellation of the "rebate", a payment which the stock exchange market pays to market makers for each transaction they consummate as a reward for the liquidity they provide.


The rebate allows Algo-Trading companies to buy and sell thousands of shares for the same rates, and profits are accumulated from stock exchange markets benefits rather than from the shares.


On the other hand, US companies which use Algo-Trading maintain that they only use publicly available information, and that their operation increases the number of transactions in the capital markets and thereby benefits the economy.


Trading behind iron doors

The Israeli Algo-Trading zone, the scope of which is revealed here for the first time, has been growing over the past three years parallel to the trend in the U.S. Legendary stories of the success of Final and similar American Algo-Trading companies encouraged Israeli entrepreneurs to form Algo-Trading start-ups, and to recruit a small team of brilliant people who are expected to develop a unique formula which can help beat the market.


Estimates indicate that over 50 such Algo-Trading companies have been formed in Israel since 2007. Estimates further indicate that approximately a half of those companies have failed and have already been wound up, or are in the process of winding up. Many of those who currently operate will not last too long. The reason for the failure is often times the same – the algorithm failed to prove its usefulness.


Like in Israeli Hi-Tech, failing entrepreneurs try time and again, while entrepreneurs who have been successful become eager to pursue further success, form new companies or help to fund them. Research conducted by Calcalist Magazine indicated that members of Final, Israel's most profitable Algo-Trading company and one of the world's most successful ones, formed no less than five new Algo-Trading companies since 2006.


Gal Kelkstein, 26, Final's founder Nir Kelkstein's younger brother, formed a company called Stock-Tech in 2007. He was assisted in doing so by Final's co-founder, Noam Stern-Perry, and Final's shareholders Rami Lipman and Attorney Yoav Ben-Dror. One of the company's investors is the father of the Kelkstein brothers, Arie Kelkstein. Stock-Tech does not compete with Final, but rather is considered the "Israeli arm" of the Kelkstein family. It is active in the Tel Aviv Stock Exchange, whereas Final is active in derivatives markets abroad.


Ben-Dror, Lipman and Shapiro helped fund another Algo-Trading company named Quantis in 2007, which focuses on derivative trading abroad. Quantis’ founder is a young entrepreneur called Tzachi Melamed, 34, who graduated two years ago with a master’s degree in finance from the University of Tel Aviv. Melamed has been working since he was young in the high-tech industry, and a few years ago he attempted to form a company that failed to break into the market.


Another shareholder in Final who was recently involved in the formation of an Algo-Trading company is Yariv Gilat, 40. Gilat, who according to sources holds approximately 10% of the shares of the cash machine of Herzliya, helped form an Algo-Trading company called Delaven, which, since its inception in 2006, has been keeping a low profile working from a private house in Ramat Hasharon, and which recorded gains of several millions of dollars in automatic trading in arbitrage shares.


Delaven was formed by Chen Munitz, 35, formerly a trader in the Gift Investment House and one of the founders of its Arbitrage Department. Gilat assisted him by, among other things, introducing him to various global contacts to whom he himself was introduced while working for Final.


Adi Yona, 33, a former worker at Final who holds a small percentage of the company's shares, left the company in 2006 and approximately a year later joined Pie Finances, an investments house engaged in trading in its nostro accounts. Yona formed and presides over the Algo-Trading operations of the company. Pie's algorithm recently started operating in the Tel Aviv Stock Exchange; however, thus far, it has only yielded insignificant gains.


Nir Kelkstein also recently turned to new enterprises in the medical sphere. Sources close to him told Calcalist Magazine that he currently invests in BSP, an Israeli medical device company which develops a program for early identification of heart attacks.


A pre-eminent figure within the Israeli Algo-Trading industry is the mathematician Dr. Shay Pilpel, 59, who had worked until recently for the successful US company Citadel. Pilpel founded the Algo-Trading company Wexford Capital in 2006, which trades in complex financial instruments and in derivatives, and its offices operate from the Sea and Sun apartment complex near the Cliff Beach in Tel Aviv.


Pilpel, a retired commander of a branch in the 8200 Intelligence Unit, is a well-known figure within the Israeli capital market, and his acquaintances say about him that "he is considered particularly brilliant, even by the standards of the environment in which he operates."


The new area also attracts people from traditional industries, as well as investors in the High-Tech industry: among Quantis’ investors are David Kite, one of the owners of British financial company Kite Capital; Brian Cooper one of the founders of the software company Retalix; and General (retired) Giora Rom. Shaul Maor of Maor Luski Investments House and Eden Shochat, a High-Tech entrepreneur who was among the founders of the start-up Shape and a partner in the American Internet Company are also among the private investors of Delaven.


Gilad Altshuler and Kalman Saham, owners of Altshuler Saham Investments House, also invest privately in Algo-Trading. The name of their company is Goldfish, and it operates from the neighborhood of State Square in Tel Aviv. It was formed in 2008 by Amnon Riftin, formerly an employee of FMR, an international company which is one of two companies marketing HFT programs for fast trading in Israel. The other company which markets HFT programs is Sibron Information Technologies. Goldfish's algorithm is used for option trading in Seoul. Sources privy to the company's operations told Calcalist Magazine that since few companies operate through HFT programs in the Far East, Goldfish, despite its youth, has managed to make impressive profits.


Other familiar faces from the capital market have entered the world of Algo-Trading in the recent years. One is Yaakov Grossman, an experienced bond and foreign exchange trader who entered the innovative industry through the Aracamba Company which he owns. According to information obtained by Calcalist Magazine, Aracamba conducts its automatic trading both of its own capital and for Bank Hapoalim, and it is considered one of the most active and profitable companies in the industry.


Other familiar names in the field are Michael Davis, known as the "first option trader in Israel", who invests in the development of his own trading algorithm; Eyal Bakshi, of the owners of Barak Capital, who engages in Algo-Trading activity in bonds in Israel and in Europe; and Moshe Barad, a former employee of Final. Barad formed the Bareket Information Systems company in 2007, which operates from Ramat Gan and recently began Algo-Trading options on the Tel Aviv 25 index.


Another intriguing character is Gadi Zohar, 40, a former Industrial Engineering professor of the Technion, who has published articles regarding derivative trading. A few years ago, he entered the Algo-Trading industry, made millions in the Israeli options market and recently relocated to London.


Two other companies which are flying under the radar but landing successfully are Leprechaun, formed by Ben Boxman, a computer engineer and alumnus of the 8200 unit. The company, despite its recent entry into Algo-trading, has been reported to have recorded substantial profits. The second company is Prolific, a company formed by programmer Yariv Eizenberg. Both companies are active in option trading in the Tel Aviv Stock Exchange.


Keeping a low profile is a mandatory requirement for owners of Algo-trading companies, many of whom do not even know their competitors and learn of them through searching through publications and small blogs from around the world. Many of the Algo-Trading entrepreneurs declined to be interviewed, and those who agreed requested to do so off the record and on a no-name basis.


Silence, they explained, is essential in order to protect the most important asset in the industry: The mathematical algorithms through which they hope to outperform the markets. If the formula reaches a competitor, its fate will be that of any investment formula that reaches the public domain – it will stop working. The competitor will use it, take a bite of the developer's profits, and shortly thereafter the computers that use it will stop recording gains.


Algorithmicans – the people who develop the formulas, based on which the computers will determine in which shares, bonds or derivatives to trade – are in high demand for employment in the industry. This is the reason that the companies’ entrepreneurs keep searching for brilliant graduates of exact science and engineering departments and of elite IDF technological units.


"Most names come from friends' recommendations", says an Israeli Algo-Trading entrepreneur who is currently recruiting, "but this doesn't guarantee anything. A doctorate in physics will be of no use if the Algorithmician does not exhibit spark and creative thinking. The demand is not for people who are good with theories or just smart people, but for people of driving intelligence".


"Moreover, it is not always easy to attract those who are suited for the job", adds an owner of another Algo-Trading company. "Many of them receive tempting and stable offers from Intel and other leading technology companies. But when you explain to them what the industry is about and how thriving it is, when you sell them the dream, it often times works."


'You became irrelevant overnight'

In a Tel Aviv tower, inside a small, windowless room, rest greenish computer screens. An endless list of data runs on these screens at a pace which is beyond the abilities of the human eye. Those are trading instructions. Four youngsters who established an Algo-Trading company in the past year are sitting in front of the modem.


“Most people fall, some survive”, say Oran, the oldest of the four, who is 30 and has more experience in the capital markets than the others. “In the meantime, our profits curve is moving in one direction, and we currently spend 15 daily hours on improving our algorithm. We are headed in the right direction, en route to generating profits independent of market performance”.


Can you understand what the computer before you does?


Shay, 25, the youngest of the four, giggles. “You need to see The Matrix”, he says.


The basic assumption of founders of Algo-Trading companies is that the capital markets really do behave like the world mirage of the “Matrix” – a world, the picture of whose reality conceals a different reality, and within it the one who can understand its rules will make superior gains. The “picture of reality” in the capital markets is that is impossible to foretell the price of a stock in the next second and the chance that it will go up is equal to the chance that it will go down. Based on such a picture, profits in daily trading in the stock exchange are the result of risk management, identifying trends, joining them and developing professional instincts.


Algo-Trading companies invest huge amounts of money in developing formulas that will discover a different reality, in which it is possible to predict whether the price of a stock will go up or down in the next second in a likelihood that is higher than 50%. Such formulas seek to find statistical links between many minor manifestations which occur every second in the capital markets. Just as investors learn about the link between hikes in the NASDAQ and in the Dow Jones, or between a fall in oil prices and rise in sugar prices, the formulas search for connections between tens of variables, and attempt to achieve a minor statistical edge in foretelling a stock price.


Other Algo-Trading formulas seek to be the first to identify manifestations such as an attempt of a major institution to purchase a large amount of shares on a piecemeal basis. Major institutions often opt for the piecemeal method to avoid stimulating demand for a stock which will result in a hike. Once the Algo-trading computer has identified a piecemeal purchase of shares, it will preempt the major institutions’ purchase of the shares and subsequently sell them at a higher rate.


In statistics, however, formulas can be wrong, and dozens of quick transactions can result in loses. Nonetheless, when the gamble is carried out endless times around the clock, the statistical advantage, even if small, is effective and profits mount. “But this doesn’t last forever” tells us an entrepreneur of a young Israeli Algo-Trading company which currently records annual profits of millions of New Israeli Shekels. “Today you’re up and all the brokers want to work with you, but tomorrow your formula could stop working and you can suddenly collapse and become irrelevant. In any event, it’s going to change the stock exchange. Businessmen who sit in coffee shops in the City of Tel Aviv and chat about technical analysis and forecasts of analysts will stay in the Stone Age. Algorithms are the future of the investment world.”


Loser: The economy

Capital market experts in the US do not view Algo-Trading favorably. Their main concern is that the system is unfair: it involves the use of a technological edge to boost profits in the markets, at the expense of investors who do not have access to fast computers with artificial intelligence.


This argument is met with approval even by some Algo-Traders. “It doesn’t matter how hard Algo-Trading companies will try to convince us that everyone has access to their information – it is just not true,” admits an Israeli entrepreneur who is active in Algo-Trading, and who, like others who agreed to interview, declined to reveal his identity. “The technology does provide us with information before the others and yes, it is unfair,” he adds.


Some economists, including Paul Wilmott, founder of the reputable Wilmott finance magazine, believe that Algo-Trading companies endanger the stability of the capital market and that their spread will result in the creation of bubbles and their burst will worsen. “This is the problem with that industry”, he wrote in his article “Hurrying into the Next Panic” which he published in late July in the New York Times. “They only care about boosting their profits and not about the resulting volatility in the stock exchange. And volatility is bad for the economy.”


In his New York Times article of early August, Nobel Laureate in Economics Paul Krugman criticized the Algo-Trading companies for generating no economic value. The purpose of the capital market, he reasoned, is to promote the economy by funding companies which are eager to make their visions come true, and the purchase of a stock and its sale a second later does nothing to contribute to this purpose.


“And the issue of pre-determined death is an added complication”, notes the owner of an Israeli Algo-Trading company, which was profitable in the past and recently dissolved after its formula had stopped working. “In this market, 90% fail right from the start, and the rest don’t survive for too long. The stock exchange learns your formula very quickly, and then places a mirror in front of your face and tells you that you are nothing”. A senior in the software company Sivron, which, together with the company FMR provides software for faster trading instructions to Algo-Trading companies says: “many experienced traders in the capital market came to me with fantasies, were sure they deserved a Noble Prize, but very quickly faded away.”


An Algo-Trading entrepreneur who vanished and agreed to reveal himself is Ronen Gil, owner of Red & Green, a company which is likely to dissolve in the near future. “When I was a student, I came up with some formula for trading in the Tel Aviv 25 and decided to exploit it,” he says. “But I soon realized, the hard way, that, to succeed in the stock exchange, it’s not enough to have market expertise and that you also need technological expertise. There was one day in which 12% of our profits were written-off within a second because of a bug in the program”.


Final’s founders also experienced this, when they formed the "Magic Box Company" in 2007, which attempted to forecast prices of commodity derivatives within hours. The formula did not prove helpful and the company dissolved by the end of 2008. But if Final’s success story happened once, so believe its founders, as well as Ronen Gil and the other entrepreneurs in the young industry, it can happen again. After all, it is just the beginning.



פרסום ראשון: 01.17.11, 19:31
 new comment
This will delete your current comment