Since the exchanges began their journey of transformation, they have faced many challenges in building a reliable infrastructure. For exchanges today, their technology has to work seamlessly. Any outage has a direct impact on the trading volume on the exchanges. An unreliable exchange could literally lose liquidity and significant market share. In addition, due to the modular exchange model and interdependence between vendors, a failure on one exchange could have an impact on other exchanges halfway around the world. Over the last two decades, a number of exchanges have faced outages lasting from a few minutes to a full day. For example, Euronext's Paris-based exchange suffered over 10 trading suspensions in 2002 alone. By comparison, DTB suffered only one outage on April 16, 2002. ¹ And in today's competitive world, these outages can be dreadful.

As one of the early entrants in electronic trading, LIFFE has had time to improve its electronic trading infrastructure. With any new launch, people expect problems to occur. Therefore, the technical issues in the early stages were easier to tolerate. There were few competitors and even fewer people in electronic trading. However, even in those early days, any outages were noticed since traders made their living trading these markets. And traders are not very patient people. So an outage of even a few minutes is bound to create many unhappy traders. In 2001 alone, LIFFE suffered more than six different outages that brought its entire trading system to a screeching halt. For example, the system was down due to technical issues on August 23, and the exchange was forced to halt trading in Gilt futures, sterling interest rate futures, Euro interest rate futures, swap futures, and U.S. stock index futures. ⁶ In December 2006 it suffered an hour-long outage that once again brought down its trading system.

In June 2007, LIFFE suffered one of its biggest outages since 1997, the year when it launched LIFFE Connect and completed its transition to electronic trading. ⁷ The exchange was forced to shut down its operations from 7:00 a.m. to 11:00 a.m. when the exchange monitoring systems broke down. The exchange officials noticed an issue with the Connect tools that it used to monitor and control the markets. The markets are automatically opened at 7:00 a.m. When the issue was identified, the exchange moved to delay the opening of the exchange; however, by the time the system halt was implemented, some trading had already occurred for some of its big products, such as Euribor, ⁸ UK Gilt, and CAC 40 Index. Trading in Euribor and UK Gilt⁹ was open for 21 seconds and almost 20 minutes for CAC 40 Index. ¹⁰ All the orders submitted for these products were canceled by the exchange. The exchange system was fully operational at 11:30 a.m. (London Time), almost four and a half hours later than its usual opening hour.

LIFFE has had a history of outages and system failures. Although the exchange issued an apology and moved to reassure its customers that the issue was identified and resolved, the exchange's reputation for system stability and reliability suffered significantly compared to LSE, which had not suffered a single outage in its seven-year-old technology, SETS. Of course, as we will now discuss, with its move to a new-generation technology platform, LSE suffered one of the worst outages in the financial markets as well.

On September 8, 2008, London traders were bracing for one of the biggest rallies in a market that had been in a downward spiral for quite some time due to the mortgage crisis, which had almost brought the global financial system to its knees. The U.S. government had stepped in to rescue Freddie Mac and Fannie Mae to help ease the tension in the financial markets. London traders were ready for a busy day. They got a glimpse of the busy trading when over 270 million shares exchanged hands in only the first hour of trading, compared to 617 million traded in the previous day. ¹¹ At approximately 9:00 a.m., LSE announced it was shutting down its trading system when it noticed connectivity failure between the exchange and several brokers. The traders were unable to connect to LSE's 15-month-old trading system, TradeElect, a proprietary trading system that promised to provide faster round-trip order execution and expanded capacity to handle the growing volume. Trading was not resumed until 4:00 p.m., 30 minutes before the official close. It left hundreds of traders on LSE out of the trading activity for almost the entire day.

On a day that would have seen some significant trading, the LSE system came to a complete standstill. For the next seven hours the traders could not trade a single product on LSE, and the exchange volume took quite a hit. Shares of bank stocks such as Barclay, HBOS, and Lloyds TSB listed on the LSE traded 30%, 31%, and 21% less than their past three-month daily average. By comparison, bank stocks listed on other markets such as UBS AG and Deutsche Bank AG saw their volume double that day compared to their past daily average volumes. The LSE traded a total of $7.5 billion worth of shares, which was only about half of its usual volume. ¹² The FTSE-100 was trading 200 points up an hour into trading when suddenly the traders could not route orders through the LSE trading platform. And for the next seven hours thousands of screens connected to the LSE trading platform were frozen with the FTSE 100 stuck at 5440.2, up 199.5. ¹³

For the LSE this was only the second major incident since 2000 that kept traders out of the market for over eight hours. The occasional system glitches are expected and reluctantly tolerated by traders, but this recent outage could not have come at a worse time. The LSE had been fighting to remain independent while other exchanges were merging to increase their market share. And new alternative trading systems such as Chi-X and Turquoise were arriving. Only a year old, these two trading platforms offered high-speed trading platforms at low cost, something the trading community, which had been increasing its use of automated trading, gravitated toward. If these system reliability issues continued on the LSE, trading platforms such as Chi-X, which already claimed to have 13% of FTSE 100 volume, ¹⁴ would most certainly pose a more formidable competitor to the LSE. Although this particular outage did not provide a significant boost to the trading volume on these other two electronic trading platforms, it was primarily due to both systems being fairly new, and few trading firms had signed on to trade on these platforms. Outages such as these made the new electronic trading platforms a possible alternative for trading firms that were virtually idle for most of this trading day.

One of the leading U.S. options exchanges, the CBOE has been late to the electronic trading party. Just like NYSE, the CBOE took a cautious approach to electronic trading. It still maintains its floor operations for options trading; however, the majority of its trading now flows through its hybrid trading system, launched in 2003. As described in a previous chapter, this system allows CBOE customers to route trades through the CBOE's hybrid system, CBOEdirect, or to continue to trade on the floor. The intention of the hybrid system is to provide an integrated market, which includes both the floor-trading quotes and electronic trading. Today over 92% of the CBOE's orders flow through the electronic system; however, 45% of the trading volume still occurs on the floor. Nevertheless, the impact on trading volume was certainly apparent on November 10, 2006, when the CBOE's electronic system suffered technical issues. The system was unavailable to the trading community from approximately 12:00 p.m. EST until the end of the trading day¹⁶ on Friday. The CBOE Futures Exchange (CFE), ¹⁷ which is fully electronic, was also unavailable for the remainder of the day. Trading for the CBOE was only available on the floor, which remained open for the entire day. Although floor-trading was available, the CBOE volumes were lower than the previous day. Approximately 1.8 million options were traded, compared to 2.8 million on the previous day. The trading firms for the U.S. options markets have alternatives. The firms using electronic trading platforms can route orders to exchanges such as the ISE or Philadelphia Options Exchange. And the competitor exchanges saw an increase in their volumes as a result of the system outage at the CBOE. However, the trading firms had no alternatives for contracts such as S&P 500, which is exclusively listed on the CBOE. Trading of the S&P 500 was significantly lower than the daily average volume. On Friday, approximately 225,000 contracts exchanged hands compared to daily average volume of 430,000. The lighter volume is certainly an indication of the increasing reliance on electronic trading by the financial community. So even though the floor was open for trading, the exchanges logged lower trading volume.

Exchanges around the world are completing their migration toward electronic trading. The trading community is getting accustomed to the availability of real-time data and the trading applications used to route orders. So far we have seen technology glitches that cause system outages and halt trading on an exchange. And we have seen the impact incorrect data has in a real-time trading environment. One of the other major impacts of technology is the potential volatility in the market that can be caused by system latency and capacity issues. As we will see, these issues are not due to actual failures in the system or inaccurate information. Instead, it's a potential delay in data dissemination that can make the market appear to be far more volatile than it is. In the automated trading world, traders build algorithms and strategies specifically to react to large volatility swings in the markets. If the volatility in the markets is caused abruptly due to capacity issues, it can trigger unwanted orders that can be costly to the trading community.

On December 13, 2005, a simple human error led to an instant $331 million loss for Mizuho Securities Company. A trader at Mizuho was preparing to sell one share of J-Com at a price of 610,000 yen. The actual order submitted was 610,000 shares of J-Com at 1 Yen! News of the bargain price traveled fast, and buyers quickly bought shares of J-Com. It was an early Christmas present for day traders, who were the most active buyers, but they were not alone. By the end of the trading day, Morgan Stanley, a large investment bank, owned 31.5% of J-Com. Although the trades were not busted, there was substantial criticism of the exchange for failing to stop the trade even after the trader realized his mistake and tried to cancel the trade. A simple human error led to frenzied trading on the Tokyo market, contributing to an almost 300-point drop on the Japanese index, the Nikkei 225, which closed down 2%. ³¹ In the end, six of the trading firms agreed to repay $141 million. Mizuho is expecting to recoup the remaining amount from the exchange due to the lack of controls on the exchange to stop erroneous trades.

Similarly, one of the traders at the former Lehman Brothers erased almost $50 billion (30 billion pounds) off the FTSE index in 2001 when he accidentally submitted £300 million for a trade instead of £3 million. The erroneous trade caused the FTSE 100 to drop almost 120 points. Another fat-finger mistake caused an almost 100-point drop in the Dow Jones Industrial Average when a trader at Bear Stearns entered a $4 billion sell order instead of $4 million. ³² These mistakes, although caused by a single firm, have widespread impact on the markets and require exchanges to deal with these trades and the potential losses and adjustments, which at times can take weeks.

The fat-finger mistakes are still likely to occur in the electronic trading world. The trading firms and exchanges continue to implement controls to catch orders that might be caused by a fat-finger mistake. For example, CME has price banding limitations, which restricts its trading platform from accepting orders outside a specified trading range for products. The front-end trading systems have a number of checks built in to control a potential incorrect order from flowing to the exchanges. These applications can set controls to limit the size and price of orders submitted by traders. For example, if a trader sets a limit for order size at 1000, an alert would pop up on his screen to confirm a trade larger than 1000. Although these controls are likely to reduce the number of fat-finger mistakes, they are still likely to occur.

The electronic trading system is credited with bringing competition, faster order execution, transparency, and anonymity to the financial markets. The exchanges as well as the financial community have reaped the rewards of the new model. Though electronic trading is transforming the financial industry, there still remain some challenges that must be recognized and, hopefully, overcome. The most significant issue is the inevitable unreliability of computer technology. Both the hardware and software are prone to failure. In particular, the software is terribly complicated and difficult to verify as correct. As the financial community gains more experience with electronic trading systems, it will learn to build more reliable systems that can mask and work around these failures. Since exchanges are using the modular exchange model, every component by every vendor must be built with the same devotion to quality for the entire system to remain bug free. The exchange is only as good as the weakest component in the trade cycle.

As long as humans are involved in any process, there will be human error. These errors can range from simple mistakes to rogue trading—activity by a trader who gets around limitations to put a firm at grave risk. Again, these mistakes can be mitigated by having better error detection on anything that requires manual interaction. It could be as simple as raising a warning whenever a trader inputs a trade beyond some value or banning flying objects in a trading room. When mistakes are made, exchanges will need to develop better resolution procedures than ad hoc decisions to cancel trades. A more difficult issue is dealing with panic. Electronic trading brings markets closer to the economic ideal of “perfect information,” but when the information is imperfect, traders panic and markets swing wildly. In a fast-moving market, it will be near impossible to tame fear and greed. Despite these issues, electronic trading solves far more problems than it creates.