In the last decade, technology has improved to such an extent that an increasing amount of trades occur without human intervention. Trading intelligence—recognizing buying and selling opportunities, looking for price discrepancies, finding market inefficiencies—can now be completely computerized. Electronic exchanges make it possible for proprietary trading systems to connect directly to trading platforms. At the end of the next decade, the majority of transactions will be performed by proprietary computerized trading systems and those traders who rely on human interfaces will be at a significant disadvantage.
In 1977, when I became a market maker on the Pacific Stock Exchange floor, I thought the opportunity to make money in options would be short-lived. I saw that each of the functions performed by the floor—order entry, posting of bids and offers, matching orders and clearing—could be easily replaced with technology. My timing was a little off––it wasn’t until 1990 that automated exchanges began to thrive. Although the DOT system on the NYSE automated part of the process, you never knew whether it was executed on the other side by a human being or not. The Hull Groum was first able to fully automate the transaction process in 1992 with trading on DTB (subsequently Eurex).
In the early days, computers weren’t even allowed on the floor of the options exchanges because that gave their operators an unfair advantage. In the 1980s, you could apply for approval to bring a computer on the floor. By the late eighties, computers were allowed right next to the pit and in the very late 1980s, traders discovered they could have a handheld computer with infrared capabilities which would allow them to calculate theoretical prices and access other useful market intelligence.
But computers on the floor and handhelds in the pit are only an intermediate step. Only the U.S. exchanges (and an isolated few non-U.S. exchanges) continue to resist a complete move to electronic trading. Understandably, floor members do not want to relinquish the time and place advantage that comes with the pit. Today, any customer can go to his Futures Commission Merchant and trade on automated markets as if he is in the pit. Although this occurs mostly in non-U.S. markets today, it will eventually happen in the United States. As one old-time futures trader told me recently, “open outcry is history, even if we haven’styet admitted it.”
In the future, most derivatives will be traded electronically, however, only active products will trade on electronic exchanges. Large trades and special illiquid products will trade through brokers over the phone, but may be crossed on an exchange. Automation increases participation in the market by creating a global pit, which has the potential of being much greater than any physical trading pit.
In 1999, when The Hull Group was acquired by Goldman Sachs Group, Inc., we were handling 30,000 transactions or more than one trade per second. More than 95 percent of those trades required no human intervention. Most of those trades represented arbitrage or relative pricing strategies. In the future, proprietary trading systems will be able to identify more directional trades with longer time horizons. By the year 2010, the majority of money managers will have completely automated their trade entry decisions.
In the electronic world, the new liquidity providers will be “smart” computers. These smart computers use sophisticated algorithms to constantly search for predetermined conditions that identify trading opportunities or protect positions. The computer has taken the way a specialist or local looks at the market—making small amounts of money on a large number of rapid trades—and sped it up a thousand-fold. These trades can be done time and time again without error.
Today, the information that is being used to build the algorithms is mostly price information—bid, offer, size of the bid, size of the offer, price movement and movement of correlated products. (Feeds from futures exchanges have historically given less information than their options and securities counterparts, but with the automation of exchanges, the amount of information has increased.) The average time the position is held is two to three days. As the computer is able to work with more information such as news or earnings, the time horizon will expand.
Smart algorithms are being developed that look at news. One system, for example, looks for announcements of stock splits. Anytime it sees the word “split,” it identifies the stock and buys a specified amount of shares. It recognizes and can act on changes in conditions a lot faster than a human can. Other systems analyze the words in a news item, predict the the next 10 minutes of price action in a market and take a position. A human trader looking at this news item would debate over where the price is going. The computer responds instantaneously, according to a predetermined set of parameters. Not only is human reaction time much slower than a computer, but labor is very expensive.
One of the key factors in improving trading performance of these computerized systems is the speed. In many operations speed specialists are the most valued technicians. A speed specialist’s job is to reduce both the time required for the machine to make a decision and the transmission time to the computerized exchange.
Years of research and development and millions of dollars go into the development of these algorithms, but once they are in place, the rewards are significant. (The cost of developing a proprietary trading system makes the seat price at an exchange look cheap.) Trading occurs at such speed that opportunities exist which cannot be seen with the naked eye. A blink of an eye is sufficient time to miss a trade or the opportunity to cancel an order. So in the very near future, if you have a mouse in your hand, you will be too late.
Blair Hull is principal of Matlock Capital, LLC, a family office.