The birth of electronic trading came from an unlikely source. On October 25 1984, the International Futures Exchange of Bermuda opened its doors for trading. The exchange was the brain child of Eugene Grummer, a former Merrill Lynch futures broker and, despite its ultimate failure, it set the precedent for the global shift to electronic trading platforms.
Grummer had wanted to set up shop in the US but was warned by the Commodity Futures Trading Commission that resistance from Chicago would inevitably result in years of delays for approval. Intex was initially formed in 1981 but it would take three years to come to market.
Initial trading was muted, partly due to the fact that it had launched with a gold futures contract in a bear market but also because of hostility from the Chicago floor traders. Legend has it that threats were made to brokers warning them not to support the exchange. And so the world’s first electronic exchange dwindled to nothing.
However, it was clear that the issue of electronic exchanges was not going to go away. The following year both the New Zealand Futures and Options Exchange and OM had launched with electronic trading platforms. It was not long before the big players wanted in on the action.
“In [the mid 1980s], the computer was beginning to have a real impact on the transmission of information,” says Leo Melamed, former chairman and current chairman emeritus of the CME. “Prior to that it was considered ‘futuristic’, but it began to be very real.”
Melamed had a vision of a computerised trading system that would enable transmission of orders electronically, replacing the existing method of a runner taking orders from a desk to the pit and back, with the customer having fills reported by phone. His vision became Globex, a Reuters-built, ‘after-hours’ electronic trading platform. It was officially proposed in 1987 and finally launched on June 25 1992.
Meanwhile, the Europeans stole a march on their US counterparts. At a meeting of a selection of Swiss industry leaders in the Bürgenstock resort above Lake Lucerne that was to become famous for the Swiss derivatives conference held there each year, a new exchange was touted.
Interestingly it was not necessarily the view of the future that formed the decision to launch the Swiss Options and Financial Futures Exchange as an electronic exchange, more the fact that an agreement could not be reached between the three major Swiss stock exchanges on whose floor the new exchange would trade on. “Federalism was at stake!” remembers Otto Nageli, the former Soffex chief executive.
On May 19 Soffex went live as the first fully integrated electronic exchange “an expensive and ambitious project” noted FOW. The launch of the platform came after a series of delays caused by back office glitches and the requirement to build a hugely complex real time system linking the exchange and the clearing accounts. The cost of the development of the two year project was estimated to be around SF100m (£40m, $65m).
On the first day 2,300 lots were traded despite a temporary closure following a “slight technical problem”. The technical problems were soon forgotten however as volumes rose to an average of 10,000 a day by September. By the end of the year, it was clear that the model had worked. Soffex was reporting ADV of over 8800 contracts and had recorded almost 1.4m trades by the end of the year.
An uphill battle
But replacing the open outcry method of trading would not be easy. Open outcry is said to have originated at Osaka’s Dojima Rice Market in 1730; in the late 1980s, conceiving the change that technology would have on established practice was impossible for many to imagine.
Jeff Gooch, the chief executive of Markitserv, worked as a junior at an audit firm in the late 1980s, auditing interest rate swaps. "As far as technology goes, there wasn’t any,” he recounts. “My job as a junior was to draw boxes on a page, with arrows for cash flows and marking how much money moved when. Another guy had some macros on an Apple Mac to put some values on the cash flows.”
In this environment, the traders of Chicago, and other exchanges, were sceptical of Melamed’s vision. The CME approached other exchanges about partnering to join the Globex system, but they were hesitant to commit.
Liffe pulled out of negotiations in May 1989, although it announced it would launch its own after-hours system, Automated Pit Trading – a system that built upon the Intex platform - which it did on November 30 that year. The system was run from the APT room, adjacent to the trading floor, which contained the new colour-screen computers with mouse-enabled APT software. The platform allowed firms to continue trading the products that had just closed in the pits.
“APT was literally a visualisation of the pit,” says Tom Punshon, head of service quality at NYSE Liffe. “It was as transparent a representation as possible. Having a small electronic trading offering as a complement to the main open outcry market suited us up to the point that it became necessary to adopt a full-blown daytime automated system through Liffe Connect.”
Having built the system for £1.6m, considered a good price, the Liffe board were rightly pleased. But far from supporting Melamed’s vision, sceptics at the exchange saw this as merely ‘keeping up with the Joneses’.
“It was cosmetic,” said one source, “It proved we could act fast to deliver technology when needed. What the CME had done was act too fast.”
Melamed acknowledges that Globex was stuck in a rut, “Although it worked famously well after hours, the real business wasn’t done after hours, it was done during the day. Therefore Globex was not full of liquidity because it couldn’t develop, it didn’t have enough business.”
In London, APT was serving its function as a placeholder for a fully automated system. Although it was used by traders, it lacked the intuitive user interfaces of modern trading systems.
As one former options dealer said, “Watching the pit was pretty mindboggling in itself; watching a little screen with icons coming in and out... it was dreadful! When options moved onto APT we would pay other people to watch the screens for us.”
Deutsche Terminboerse (DTB), the online German futures and options market, had launched on January 29 1990, using a system designed by Soffex, the Swiss derivatives market and adapted by Andersen Consulting, now Accenture.
Joerg Franke, who would become chief executive, later wrote that the most polite title given to him approximated to “casino owner” and that to bring an electronic system onto the market was seen by some as “the devil’s work”.
The idea of the electronic-only market met with some resistance; the resilience of its technology was questioned, as was the lack of a trading floor. By the mid-1990s, following a merger with Deutsche Börse and a failed partnership with the French derivatives market, Matif, it had attracted a number of French firms to trade on its market electronically.
In 1995, DTB overtook Matif to the number two spot in Europe. Then, in 1997, DTB launched an all-out offensive against Liffe’s dominance of the Bund Future (see page 32), which had launched on the London market in 1988.
“They incentivised brokers and market makers to start executing their business on DTB,” explains Graeme Neilly, director global account management of trading systems provider Patsystems. “They gave you a terminal for free and then a cash incentive as well.”
“To attract business the DTB cut the clearing rates,” says David Kyte, who started his own futures commission merchants (FCM), Kyte Group in 1985. “To trade a Liffe product clearing was about 56p a side and the DTB made their prices about 25p or the equivalent in Deutsche marks. It was attractive to the banks trading millions and millions each year.”
Traders warned the Liffe board that they would have to be competitive and cut prices.
“The liquidity moved out of the Liffe Bund Futures pit in two or three days,” says Neilly. “It was incredibly successful.”
Off the floor
Any doubt about the ability of electronic trading to surpass floor trading had been swept away. With its offering and cost proposition, Eurex, as DTB had become following its merger with Soffex in 1998, took the lead as Europe’s premier derivatives exchange. This showed Liffe that the market had indeed turned.
Liffe Connect was introduced in 1994 as a response to the emergence of the Bund on DTB.As an electronic trading platform it was intended to reduce the significant overheads of a trading floor and to allow the launch of new products quickly.
“We wanted to take all the advantages of a pit and put it into an automated solution with access and convenience from which ever location the trader deemed fit,” says Punshon.
The CME now faced a dilemma; it looked as though Liffe and Eurex were going to steal its lunch, something Melamed believed they were capable of. A referendum was held to allow electronic trading to be run simultaneously with open outcry in business hours.
“Electronic trading was cheaper, safer and less prone to errors,” says Melamed. “We wouldn’t prohibit open outcry but we would let the world decide which way to go, which the world promptly did.”
Liffe became fully electronic in 2000, with a model that allowed traders to choose the software they used. As it did a host of independent software vendors were waiting in the wings to offer trading systems that would give one trader an advantage over another.
“One of the guys who was ahead of the curve was Tom Theys, who used to clear through us,” says Kyte. “In the early 90s he realised if you could trade in the pit just by pressing buttons, using a device like a big Blackberry or a small computer, and you could trade directly into the pit rather than using hand signals, it would be quicker. That was the precursor for Patsystems, one of the first trading platforms.”
In the early days of electronic trading reliability was a significant issue for technology on all sides recalls Steve Carl, director of business change at NYSE Liffe, “Going a month without an outage was treated as some sort of achievement.”
Electronic trading was in its infancy, the systems were fragile and if an error occurred, freezing the system and pulling all the orders was the safest way to restore the market.
“The emphasis moved to speed and reliability but that didn’t come together to start with,” says Kyte. “There were a lot of failures; systems would go down, connectivity would go.”
Tools of the trade
Neilly says that when he joined Patsystems in June 2000, the firm offered trading software with connectivity to six markets, which grew to 60, and the advantage of electronic trading over the pit was clear.
“A big sell-side firm could have a presence on the floor and a big client might have a direct line into that firm, but if you were a retail broker you would probably have to have four or five conversations to get your order into the pit, so p0rice discovery is nigh-on impossible,” he says. “If you’re a retail broker trading the FTSE online you see the same price and volume as a trader on a prop desk.”
Patsystems succeeded partly because it used a light front end that facilitated trading over the internet and partly because of its reach. The inclusion of built-in, pre-trade risk management was a selling point as it gave the trading desks at the banks and FCMs more rigour, which the compliance teams supported.
“There was a huge fear factor at the FCMs as they started throwing screens out to traders about how they controlled that order flow. We provided them with a single system through which to do that,” says Neilly.
Ffastfill, was another ISV, founded in 2000 by a former Liffe trader, Hamish Raw. From the start it was functionally effective and provided connectivity to the electronic exchanges, enabling a trader to place orders into the market on his screen, with fat-finger and limited lot-size controls to provide risk management.
In 2003 Keith Todd, executive chairman, joined the firm, leading a technologist team, to take the company forward as a software as a service company and giving the system the scalability necessary to cope with the growing market data volumes and processing speeds.
New service oriented architectures were becoming available that enabled software to be built for service, flexibility and on a distributed basis, separating data from business logic, and putting in place sophisticated failover and monitoring capabilities into software to show how it was performing.
“In 2003 it was the early days of Microsoft.net and we adopted its toolkits but built in our own proprietary message bus,” says Todd. “For 40 years companies didn’t run the software that they built, their customers did. Now, we get to run it, monitor and improve its performance, as our clients use it.”
Distributed architectures also played their role in improving the exchange’s trading platforms. In 2005 Liffe moved to Connect v.8, switching from a centralised architecture with Enterprise class servers that were relatively large, expensive and complicated, to a distributed model which used multiple low-cost servers running Linux operating systems.
“Although hardware suppliers offered fault tolerant systems promising continuous uptime, if your application software suffered an issue in one area it could shut you down completely,” says Carl. “Using a distributed model, with application code developed to work on Linux operating systems, we entered a period of much greater reliability and significantly improved performance.”
Calm before the storm
By 2005, electronic trading had won out on both sides of the Atlantic. The CME, which had trouble processing a million trades a year prior to the implementation of Globex, could now process 10 million a day. However the rapid change was creating new challenges.
“Technology created new risks by adding complexity to the market,” says Todd. “At a base level it has improved the risk management of a basic trade. But as trading is complicated, integrating many more exchanges and types of instruments, it has increased the inherent risks and interdependency.”
Trading volumes had increased but the trades were leaving a mountain of paperwork to be done in the back office.
“We identified the need for repositories, automated settlement, confirms, but adoption was slow,” says Gooch, “Before the Fed got involved in 2005, we were still at 20-30% automation in the market.”
However the pressure from regulators was not enough. Post-trade processing did not develop prior to the collapse of Lehman Brothers in 2008, and the subsequent crisis lead regulators to mandate the automation of infrastructure.
Mifid II and the European Markets and Infrastructure regulation, along with the Dodd-Frank Act in the US, are now driving automation of the markets by legislating for standardisation of contracts and electronic trade processing. The use of central counterparties and electronic trading platforms will lead to far more efficient markets facilitating greater volumes of algorithmic trading.
The trading technology is just over a decade old and far from mature; weaknesses still appear. On October 31 the collapse of MF Global, a second tier broker, left many buy-side side firms, reliant on MF Global’s trading system, without access to the market. What these events teach the market time and again, observes Kyte, is the value of the human trader.
“There was a move towards algorithmic trading in the derivatives market, but a lot of firms have come unstuck over the last couple of years,” he says. “Failures have led firms to revert to ‘old-fashioned’ traders.”
More from FOW's 30 Year Anniversary Issue
FOW 30 Year Anniversary: foreword by Sir Brian Williamson
FOW 30 Year Anniversary: Top 30 most influential
FOW 30 Year Anniversary: 30 years of futures and options
FOW 30 Year Anniversary: 30 years of technology
FOW 30 Year Anniversary: The origins of Liffe
FOW 30 Year Anniversary: The 1987 crash
FOW 30 Year Anniversary: Bye bye bund
FOW 30 Year Anniversary: CME battles for CBOT
FOW 30 Year Anniversary: LCH after Lehmans Brothers collapsed