GLOBAL TRADING UTILIZING NEURAL NETWORKS: A SYNERGISTIC APPROACH
Chapter 5: Global Trading Utilizing Neural Networks: A Synergistic Approach
By Lou Mendelsohn,
President Market Technologies
In the late 1980s, the investment industry underwent a technology-driven revolution, which has brought about the emergence of interrelated global financial markets and the need for a global perspective on trading. Today, the world’s financial markets have become interdependent through a technological transcendence of time and space.
This trend toward globalized, interconnected markets has resulted from the confluence of two factors: advancements in satellite telecommunication and computer technologies, which can be viewed broadly as the “information technology revolution,” and the emergence of derivative financial instruments. Already, the world’s economies and financial markets have become irreversibly linked on a scale never before seen in the history of economic affairs.
Other Factors Contribute To Market Globalization
Other related factors have also contributed to the acceleration of this process. These include the proliferation of new trading exchanges and markets in emerging growth regions of the world, particularly the Pacific Rim, Asia, and South America; the declining influence of the G7 western nation central banks in the control of interest and exchange rates; international corporate consolidation through mergers and acquisitions; increased global import/export trade; recent agreements to create regional trading blocks; productivity differentials between countries; the breakup of the Soviet Union and the expansion of trade with former Soviet block nations; the globalization of corporate financing through concurrent listing of shares on several exchanges across national boundaries; and the emergence of multinational companies and financial institutions that require worldwide currency and interest-rate risk management around the clock.
The Result: One Global Financial Market
These fundamental changes in the macroeconomic dynamics involving world trade and investment capital movements across currencies, national boundaries, and time zones are ushering in an era in which there will be just one global financial market, with hundreds or thousands of interrelated component sub-markets representing the gamut of actively traded investment instruments worldwide. If this scenario unfolds as expected, it would not be far-fetched to envision a futuristic global financial system without physical trading exchanges, in which trading is conducted solely by electronic means.
The globalization of capital markets has irrevocably changed their character and nature, and is now putting previously accepted methods of financial analysis to the test.
Technical Analysis Redefined
These unprecedented changes in the financial markets necessitate a complete re-evaluation of how technical analysis and portfolio management should be performed. As the process of globalization continues to unfold, new methods of analysis that encompass the intermarket context of global trading will be needed by traders to maintain a competitive advantage. It is already a risky indulgence to adhere to an analytic perspective that focuses internally on single markets or looks only at simple linear correlations between related domestic markets. It is now mandatory to have a global investment perspective and access to the appropriate analytic tools necessary to implement it.
Information Technology Revolution
Since the early 1980s, practically no one involved in trading and financial market analysis has failed to notice the influence of computer-generated information on the financial markets. In little more than a decade, the cost of hardware, software, and market information has declined drastically, while computing, telecommunications, and software sophistication have increased dramatically.
Advances in satellite, cellular, and fiber-optic networks now allow for nearly instantaneous worldwide communication and data transmission along the emerging information, or “electronic,” superhighway. As analysis of global intermarket data becomes more widespread, investors who continue to focus narrowly on a single market’s past price data for clues to its future price direction, or maintain a portfolio of just two or three domestic asset classes, will be at a severe competitive disadvantage.
The Emergence Of Derivatives And New Markets
Derivative financial instruments are linked to the value of underlying securities. The most common derivatives include exchange-listed futures contracts, forward contracts, and options. They are used in risk management to transfer risk between two or more parties. Derivatives trading became feasible as telecommunication and computer advancements facilitated on-line linkages between exchanges trading derivatives and those trading the underlying instruments. The most talked-about example of how the marriage of information technology with derivatives trading has helped to bring about global market interdependence involved the role of program trading during the stock market crash of 1987. For all intents and purposes, this event was the first instance of a worldwide financial market phenomenon. In program trading a basket of stocks and corresponding stock index futures contracts are bought and sold simultaneously to arbitrage short-lived price differentials. Following the 1987 crash, many industry professionals attributed its abruptness and severity to program trading, citing its downward spiraling influence on the markets. Even today, there is considerable concern over the potential deleterious effects of derivatives on market liquidity during times of crisis.
Technological advancements further facilitated the expansion of derivative trading by allowing for intensive collection and analysis of market data needed to evaluate and implement derivative-related trading strategies. Until a few years ago, trading in stock indexes, futures, and options was considered too risky by many institutional investors. Those who did trade in these types of instruments did so in isolation within separate time zones on domestic exchanges. Now, the world’s derivative markets are linked by computer networks and after-hours order matching services into an around-the-clock global market. Derivatives trading is now conducted worldwide by most major financial institutions, including pension and mutual funds.
Yet, few traders comprehend the intricacies of derivatives. Little is known about how they relate to other markets, how they affect the underlying cash markets, and what might happen during an extreme market rout in which the trading system’s host computer, the trading instrument, the telecommunications network, and the counter-parties might all reside in different countries. Since most derivatives did not exist during the last major bear equities market in 1974, the degree of influence that derivatives might have in precipitating or accelerating a major worldwide financial crisis, more severe than 1987, can not yet be measured.
In fact, the ramifications of this trend toward globalized world markets are still not understood by most multinational corporate treasurers, economists, politicians, and traders. This is due primarily to the fact that global equity markets have been in a prolonged bull market since the early 1980s. With the exception of the 1987 crash and several lesser aftershocks, the financial markets have simply not been put to the acid test of illiquidity-liquidity that would occur following the onset of a major worldwide bear equities market or bond market rout.
As global integration of the world’s financial markets has accelerated, it has become necessary for traders to pay close attention to related financial markets and implement methods of analysis that take these intermarket relationships into consideration. This broader perspective is now critical to successful trading.
The interrelated markets of the 1990s offer unprecedented trading opportunities. Single-market technical analysis is no longer state-of-the-art trading technology. This method of analysis is based on the premise that a specific market’s price dynamics can be modeled sufficiently through the use of fixed trading rules and past data pertaining to that market alone. Too often, such trading rules fail to discern the intermarket and fundamental forces, or market synergy, that drives today’s global markets. This is especially true when the rules are based solely on one expert’s research into market dynamics. Also, fixed, rule-based approaches by their very nature lack the flexibility and adaptability needed in today’s volatile and rapidly changing markets.
With the potential for risk reduction and performance enhancement that can be achieved through global trading, a narrow single-market technical approach or portfolio restrictions that limit investing to domestic instruments are simply no longer reasonable. Now, dynamically adaptive analytic methods, capable of finding hidden patterns and relationships in global market data, are a sine qua non to identifying and taking advantage of global trading opportunities.
In an effort to maintain their competitive advantage in today’s world markets, institutions and sophisticated traders have begun to apply advanced computational modeling tools, such as neural networks, to the nonlinear domain of global financial forecasting and trading.
The Need For Synergistic Market Analysis
Of course, simply throwing money at new technologies is not the answer. Neither is marrying new technologies to outdated methods of analysis. Historically, two distinct schools of analysis, fundamental and technical, have been pursued. Fundamental analysis looks at basic economic supply-and-demand factors underlying the markets. Technical analysis looks internally at single markets to interpret the movement and behavior of prices as a guide to decision making. More recently, intermarket analysis, which looks intuitively at relationships between markets — usually through the subjective examination of price charts — has become fashionable. Yet, none of these approaches by itself is sufficient in today’s global markets.
In order to meet the trading challenge of the 1990s, an entirely new method of global market analysis is required. Yesterday’s lagging approaches must give way to tomorrow’s leading approaches, and subjective assessments of intermarket relationships must be supplanted by more quantitative means. This new method must recognize the non-linearity, interdependence, and interrelatedness of today’s financial markets. Most importantly, through its use, traders and investors must be able to take advantage of these conditions for profit.
In effect, nothing short of a broadened redefinition of technical analysis is in order. This new method of analysis, referred to as Synergistic Market Analysis (SMA), encompasses the more narrowly defined extant schools of technical, fundamental, and intermarket analysis (See Figure 1). This synergistic approach benefits from the use of artificial intelligence technologies, and other appropriate mathematical tools. Through their use, nonlinear relationships and complex patterns between related global markets can be quantified, thereby capturing information reflecting the intermarket dynamics, or market synergy, inherent in today’s global markets. Synergistic analysis builds on these limited methods of analysis and carries early efforts at intermarket analysis to their logical conclusion.
Synergistic Analysis Background
In 1983 — when I first published articles introducing the concept of historical simulation and back-testing in microcomputer software, and developed ProfitTaker, the first commercial software with this capability — technical analysis tools were quite primitive. Very quickly, computerized traders began using historical testing to develop trading models in which technical indicators, parameter values, and rules could be varied and “optimized” for each market. Before long, activities such as using hand-held calculators to compute various technical studies, drawing trend lines by hand on price charts, and paper-trading were replaced by software that could implement these functions more accurately and efficiently. By the mid-1980s single-market historical testing had become the backbone of computerized technical analysis, and an entire software industry was born. Yet, after considering the likely impact that market globalization would have on technical analysis, it was apparent that single-market analysis alone could not meet the demands that would be made on traders by the exigencies of global trading. More robust analytics would be needed to stay competitive.
In 1987, I developed a software program that used a spreadsheet format to correlate price movements in inter-market and expectations of impending economic reports with the price directions of various related financial markets. However, the analytic tools being used at that time in the financial industry could only indicate whether these relationships were direct or inverse. At this same time, other technical analysts also began exploring intermarket relationships, most notably John Murphy, who has since authored an excellent book on the subject, entitled Intermarket Technical Analysis.
Robust, quantitative tools needed to integrate inputs from related markets remained elusive until I began researching and experimenting with various artificial intelligence technologies. One of these, neural networks, was well suited to amalgamate technical, fundamental, and intermarket analysis. The synthesis of these three approaches was accomplished by creating one coherent analytic framework that used the computational modeling capabilities of neural networks to find nonlinear patterns and complex relationships in otherwise disparate market data.
The application of artificial intelligence, and more specifically neural networks, to financial forecasting and analysis has become a hot topic within the financial industry. Literally dozens of articles have been written on the subject in trade publications, and several books have been published in just the past few years. With an extensive amount of data readily available for analysis, neural networks are ideally suited to implement SMA by finding patterns and quantifying relationships between interrelated markets.
Prominent money management firms in various segments of the financial industry have recently announced their adoption of neural networks for pattern recognition and forecasting, hailing them as the next generation of analysis tools. However, most traders and investors, still content with single-market analysis, still only give lip service to market globalization, the need to analyze intermarket relationships, and the value of using quantitative nonlinear technologies such as neural networks.
History is replete with accounts of new technologies which, while dismissed initially by shortsighted observers, have subsequently played critical roles in redefining entire industries. Transportation via aviation, communications by telephone, and document reproduction by xerography, to name just three obvious instances, illustrate what happens when emerging technologies threaten existing ways of conducting business as usual. Those who currently dismiss global intermarket analysis through the application of nonlinear technologies will look back just a few years from now in astonishment at their own lack of foresight. Technological progress can not be stopped or even slowed down by doubters. Early adopters of nonlinear modeling capabilities to global trading will reap the financial benefits of their foresight.