Главная » Периодические издания » Subjective methods definition

1 2 3 4 ... 16

subjective methods definition

During the past several years, I have been on a quest to understand how the markets actually work. This quest has led me to researching almost every type of analysis. My investigation covered both subjective and objective forms of technical analysis-for example, intermarket analysis, Elliott Wave, cycle analysis, and the more exotic methods, such as neural networks and fuzzy logic. This book contains the results of my research.

My goal was to discover mechanical methods that could perform as well as the top traders in the world. For example, there are technologies for trading using trend following, which significantly outperform the legendary Turtle system. This book will show you dozens of trading systems and filters that can increase your trading returns by 200 to 300 percent. I have collected into this volume the best technologies that I have discovered. This overview of the books contents will give you the flavor of what you will be learning.

Chapter 1 shows how to use intermarket analysis as a predictive tool. The chapter first reviews the basics of intermarket analysis and then, using a chartists approach, explains the many different intermarket relationships that are predictive of stocks, bonds, and commodities. That background is used to develop fully mechanical systems for a variety of markets, to show the predictive power of intermarket analysis. These markets include the S&P500, T-Bonds, crude oil, gold, currencies, and more. Most of these systems are as profitable as some commercial systems costing thousands of dollars. For example, several T-Bond trading systems have averaged over $10,000 a year during the analysis period.

Chapter 2 discusses seasonal trading, including day-of-the-week, monthly, and annual effects. You will learn how to judge the reliability



2 Introduction

Introduction 3

of a seasonal trade and how to develop reliable and profitable seasonal indexes. Several winning methods for using seasonality were developed using a walk forward approach in which the seasonal is calculated only using prior data for trading stocks, bonds, and corn. This means that these results are more realistic than the standard seasonal research normally available and are very profitable. The chapter also discusses several issues relating to rhe proper use of seasonality. For example, in some markets, such as corn or other commodities that are grown, all of the available data should be used to calculate a seasonal. In markets like T-Bonds, where seasonal forces are influenced by the release of government reports, only the past N years are used because these dates change over time. Finally, several new seasonal measures are presented, beginning with the Ruggiero/Barna Seasonal Index. This new indicator combines the win percentage (Win%) and returns into one standardized measure that outperforms standard ways of selecting which seasonal patterns to trade. For example, 71 percent of our trades can be won by using the Ruggiero/Barna Seasonal Index to trade T-Bonds using walk forward analysis. Next, two other new indicators are explained: (1) seasonal volatility and (2) the seasonal trend index based on the trading day of the year. The seasonal volatility measure is valuable for setting stops: for example, when seasonal volatility is high, wider stops can be used, and when it is low, tighter stops can be used. This measure is also good for trading options, that is, for selling at premium when seasonal volatility is falling. I use my seasonal trend index to filter any trend-following system. The power of this seasonal trend index was exhibited when it predicted the trend in T-Bonds starting in mid-February of 1996. By taking the downside breakout in T-Bonds during that month, when our seasonal trend indicator was crossing above 30, I caught a short signal worth about $9,000.00 per contract in only a month and about $13,000.00 in only eight weeks.

Chapter 3 shows how fundamental factors such as inflation, consumer confidence, and unemployment can be used to predict trends in both interest rates and stocks. For example, one market timing model has been 90 percent accurate since August 1944, and would have produced better than the standard 10 to 12 percent produced by buy and hold and was in the market about half the time.

Chapter 4 discusses traditional technical analysis, beginning with why some people say technical analysis does not work and why they are wrong. Several powerful trading strategies based on technical analysis are used

by professional traders to exploit inefficiencies in the markets and make money. These strategies range from position to day trading.

Chapter 5 explains what the commitment of traders (COT) report is, how it is reported, and how to use it to develop market timing models. Several system examples are provided.

Chapter 6 is an overview of how general statistical analysis can be applied to trading. To make you a more profitable trader, the following statistical measures are discussed:

Mean, median, and mode

Types of distributions and their properties

Variance and standard deviation.

Interrelation of gaussian distribution, mean, and standard deviation. Statistical tests that are of value to trading system developers Correlation analysis.

This chapter serves as a background to much of the rest of the book.

Chapter 7 first explains the nature of cycles and how they relate to real-world markets. Later, you will see how cycles can be used to develop actual trading strategies using the maximum entropy method (MEM), or maximum entropy spectral analysis. MEM can be used to detect whether a market is currently trending, or cycling, or is in a consolidation mode. Most important, cycles allow discovery of these modes early enough to be of value for trading. A new breakout system, called adaptive channel breakout, actually adapts to changing market conditions and can therefore be used to trade almost any market. During the period from 1/1/80 to 9/20/96, this system produced over $160,000.00 on the Yen with a drawdown of about $8,700.00. Finally, the chapter tells how MEM can be used to predict turning points in any market.

Chapter 8 shows how combining statistics and intermarket analysis can create a new class of predictive trading technology. First, there is a revisit to the intermarket work in Chapter 1, to show how using Pearsons correlation can significantly improve the performance of an intermarket-based system. Several trading system examples are provided, including systems for trading the S&P500, T-Bonds, and crude oil. Some of the systems in this chapter are as good as the high-priced commercial systems.

The chapter also discusses a new indicator, predictive correlation, which actually tells how reliable a given intermarket relationship currently is



when predicting future market direction. This method can often cut drawdown by 25 to 50 percent and increase the percentage of winning trades. Intermarket analysis can be used to predict when a market will have a major trend. This method is also good at detecting runaway bull or bear markets before they happen.

Chapter 9 shows how to use the current and past performance of a given system to set intelligent exit stops and calculate the risk level of a given trade. This involves studying adverse movement on both winning and losing trades and then finding relationships that allow setting an optimal level for a stop.

In Chapter 10, system control concept feedback is used to improve the reliability and performance of an existing trading strategy. You will learn how feedback can help mechanical trading systems and how to measure system performance for use in a feedback model. An example shows the use of a systems equity curve and feedback to improve system performance by cutting drawdown by almost 50 percent while increasing the average trade by 84 percent. This technology is little known to traders but is one of the most powerful technologies for improving system performance. The technology can also be used to detect when a system is no longer tradable-before the losses begin to accumulate.

Chapter 11 teaches the basics of many different advanced technologies, such as neural networks, machine induction, genetic algorithms, statistical pattern recognition, and fuzzy logic. You will learn why each of these technologies can be important to traders.

The next three chapters tell how to make subjective analysis mechanical. Chapter 12 overviews making subjective methods mechanical. In Chapter 13, I explain Tom Josephs work, based on how to identify mechanical Elliott Wave counts. Actual code in TradeStations EasyLanguage is included. In Chapter 14, I develop autorecognition software for identifying candlestick patterns. A code for many of the most popular formations, in EasyLanguage, is supplied.

The next topic is trading system development and testing. Chapter 15, on how to develop a reliable trading system, will walk you through the development of a trading system from concept to implementation. Chapter 16 then shows how to test, evaluate, and trade the system that has been developed.

hi the final chapters, I combine what has. been presented earlier with advanced methods, such as neural networks and genetic algorithms, to develop trading strategies.

Chapter 17 discusses data preprocessing, which is used to develop models that require advanced technologies, such as neural networks. The chapter explains how to transform data so that a modeling method (e.g., neural networks) can extract hidden relationships-those that normally cannot be seen. Many times, the outputs of these models need to be processed in order to extract what the model has learned. This is called postprocessing.

What is learned in Chapter 17 is applied in the next three chapters. Chapter 18 shows how to develop market timing models using neural networks and includes a fully disclosed real example for predicting the S&P500. The example builds on many of the concepts presented in earlier chapters, and it shows how to transform rule-based systems into supercharged neural network models.

Chapter 19 discusses how machine learning can be used to develop trading rules. These rules assist in developing trading systems, selecting inputs for a neural network, selecting between systems, or developing consensus forecasts. The rules can also be used to indicate when a model developed by another method will be right or wrong. Machine learning is a very exciting area of research in trading system development.

Chapter 20 explains how to use genetic algorithms in a variety of financial applications:

Developing trading rules

Switching between systems or developing consensus forecasts. Choosing money management applications. Evolving a neural network.

The key advantage of genetic algorithms is that they allow traders to build in expertise for selecting their solutions. The other methods presented in this book do not offer this feature. Following a discussion of how to develop these applications, there is an example of the evolution of a trading system using TSEvolve, an add-in for TradeStation, which links genetic algorithms to EasyLanguage. This example combines intermarket analysis and standard technical indicators to develop patterns for T-Bond market trades.



CLASSICAL MARKET PREDICTION

Part One

Classical Intermarket Analysis as a Predictive Tool

WHAT IS INTERMARKET ANALYSIS?

Intermarket analysis is the study of how markets interrelate. It is valuable as a tool that can be used to confirm signals given by classical technical analysis as well as to predict future market direction. John J. Murphy, CNBCs technical analyst and the author of Intermarket Technical Analysis (John Wiley & Sons, 1991), is considered the father of this form of analysis. In his book, Murphy analyzes the period around the stock market crash of October 19, 1987, and shows how intermarket analysis warned of impending disaster, months before the crash. Lets examine some of the intermarket forces that led to the 1987 stock market crash.

Figure 1.1 shows how T-Bonds began to collapse in April 1987, while stocks rallied until late August 1987. The collapse in the T-Bond market was a warning that the S&P500 was an accident waiting to happen; normally, the S&P500 and T-Bond prices are positively correlated. Many institutions use the yield on the 30-year Treasury and the earnings per share on the S&P500 to estimate a fair trading value for the S&P500. This value is used for their asset allocation models.

T-Bonds and the S&P500 bottomed together on October 19, 1987, as shown in Figure 1.2. After that, both T-Bonds and the S&P500 moved in a trading range for several months. Notice that T-Bonds rallied on the




FIGURE 1.1 The S&P500 versus T-Bonds from late December 1986 to mid-September 1987. Note how stocks and T-Bonds diverged before the crash.


FIGURE 1.2 The S&P500 versus T-Bonds from mid-September 1987 to early May 1988. T-Bonds bottomed on Black Monday, October 19, 1987.

day of the crash. This was because T-Bonds were used as a flight to

safety.

T-Bond yields are very strongly correlated to inflation; historically, they are about 3 percent, on average, over the Consumer Price Index (CPI). Movements in the Commodity Research Bureau (CRB) listings are normally reflected in the CPI within a few months. In 1987, the CRB had a bullish breakout, which was linked to the collapse in the T-Bond market. This is shown in Figure 1.3. The CRB, a basket of 21 commodities, is normally negatively correlated to T-Bonds. There are two different CRB indexes: (1) the spot index, composed of cash prices, and (2) the CRB futures index, composed of futures prices. One of the main differences between the CRB spot and futures index is that the spot index is more influenced by raw industrial materials.

Eurodollars, a measure of short-term interest rates, are positively correlated to T-Bonds and usually will lead T-Bonds at turning points. Figure 1.4 shows how a breakdown in Eurodollars preceded a breakdown in T-Bonds early in 1987.


FIGURE 1.3 T-Bonds versus the CRB from October 1986 to June 1987. The bullish breakout in the CRB in late March 1987 led to the collapse in the T-Bond market in April 1987.




FIGURE 1.4 T-Bonds versus the Eurodollar for the period September 1986 to May 1987. The breakdown in Eurodollars in late January 1987 preceded the collapse in the T-Bond market in April 1987.

Figure 1.5 shows how the gold market began to accelerate to the upside just as Eurodollars began to collapse. Gold anticipates inflation and is usually negatively correlated with interest-rate-based market rates such as the Eurodollar.

Analysis of the period around the crash of 1987 is valuable because many relationships became exaggerated during this period and are easier to detect. Just as a total solar eclipse is valuable to astronomers, technical analysts can learn a lot by studying the periods around major market events.

Given mis understanding of the link between the S&P500 and T-Bonds, based on the lessons learned during the crash of 1987, we will now discuss intermarket analysis for the S&P500 and T-Bonds in more detail.

Figure 1.6 shows that T-Bonds peaked in October 1993, but the S&P500 did not peak until February 1994. The collapse of the bond market in early 1994 was linked to the major correction in the S&P500, during late March.



FIGURE 1.6 The S&P500 versus T-Bonds for the period August 1993 to April 1994. The 1994 bear market in T-Bonds led to the late March correction in the S&P500.



T-Bonds continued to drop until November 1994. During this time, the S&P500 was in a trading range. The S&P500 set new highs in February 1995 after T-Bonds had rallied over six points from their lows. This activity is shown in Figure 1.7.

Figure 1.8 shows the Eurodollar collapse very early in 1994. This collapse led to a correction in the stock market about two weeks later. This correction was the only correction of more than 5 percent during all of

1994 and 1995.

Figure 1.9 shows that the Dow Jones Utility Average (DJUA) also led the S&P500 at major tops. The utilities topped in September 1993-a month before bonds and five months before stocks.

Figure 1.10 shows that the S&P500 and DJUA both bottomed together in November 1994.

With this background in intermarket relationships for the S&P500, lets now discuss the T-Bond market.


FIGURE 1.7 The S&P500 versus T-Bonds for the period September 1994 to May 1995. When T-Bonds bottomed in November 1994, stocks did not break the February 1994 highs until February 1995.


FIGURE 1.8 The S&P500 versus Eurodollars for the period September 1993 to May 1994. The collapse in Eurodollars was linked to the late March 1994 correction in the stock market.


FIGURE 1.9 The S&P500 versus the Dow Jones Utility Average for the period July 1993 to March 1994. The DJUA peaked in September 1993. Stocks did not peak until February 1994.



Classical Market Prediction


FIGURE 1.10 The S&P500 versus the Dow Jones Utility Average for the period August 1994 to April 1995. The S&P500 and the DJUA bottomed together in November 1994 and rallied together in 1995.

Figure 1.11 shows that the Dow Jones Utility Average (DJUA) led the bond market at the top during several weeks in late 1993. The DJUA is made up of two different components: (1) electrical utilities and (2) gas utilities. Before T-Bonds turned upward in late 1994, the electrical utilities began to rally. This rally was not seen in the DJUA because the gas utilities were in a downtrend. This point was made during the third quarter of 1994 by John Murphy, on CNBCs Tech Talk. Figure 1.12 shows how the electrical utilities are correlated to T-Bond future prices.

One of the most important things that a trader would like to know is whether a current rally is just a correction in a bear market. The Dow 20 Bond Index, by continuing to make lower highs during 1994, showed that the rally attempts in the T-Bond market were just corrections in a bear market. This is shown in Figure 1.13. The Dow 20 Bond Index is predictive of future T-Bond movements, but it has lost some of its predictive power for T-Bonds because it includes some corporate bonds


FIGURE 1.11 The T-Bond market versus the Down Jones Utility Average. The DJUA peaked a few weeks before T-Bonds in late 1993


FIGURE 1.12 T-Bonds versus the Philadelphia electrical utility average for the period August 1994 to February 1995. The electrical average turned up before T-Bonds in late 1994.




FIGURE 1.13 T-Bonds versus the Dow 20 Bond Index for the period March 1994 to November 1994. The Dow 20 Bond Index is in a downtrend, and short-term breakouts to the upside fail in the T-Bond market.

that are convertible to stock. This property also makes the Dow 20 Bond Index a very good stock market timing tool.

Copper is inversely correlated to T-Bonds, as shown in Figure 1.14. The chart shows that copper bottomed in late 1993, just as the T-Bond market topped. The copper-T-Bond relationship is very stable and reliable; in fact, copper is a more reliable measure of economic activity than the CRB index.

Many other markets have an effect on T-Bonds. One of the most important markets is the lumber market. Lumber is another measure of the strength of the economy. Figure 1.15 shows how T-Bonds have an inverse relationship to lumber and how lumber is predictive of T-Bonds.

Crude oil prices, another measure of inflation, are inversely correlated to both T-Bonds and the Dollar index. The inverse correlation of crude oil and T-Bonds is depicted in Figure 1.16.


FIGURE 1.14 T-Bonds versus high-grade copper. Copper bottomed in late 1993 just as T-Bonds topped.


FIGURE 1.15 T-Bonds versus lumber from late September 1995 to the end of March 1996. Lumber was in a downtrend during late 1995 while T-Bonds were rising.




FIGURE 1.16 T-Bonds versus crude oil. In general, T-Bonds and crude oil have a negative relationship.

Many other markets are predictive of T-Bonds. For example, many of the S&P500 stock groups have strong positive or negative correlation to T-Bond prices. Some of these groups and their relationships are shown in

Table 1.1.

We will now discuss the Dollar, which is normally negatively correlated with the CRB and gold. Figure 1.17 shows that the breakout in gold in early 1993 led to a double top in the Dollar. Later, when gold and the CRB stabilized at higher levels, the Dollar had a major decline, as shown in Figure 1.17.

Lets now look at foreign currencies, The Deutsche Mark (D-Mark) was in a major downtrend during 1988 and 1989 and so was Comex gold. The D-Mark and gold broke out of the downtrend at the same time, as shown in Figure 1.18.

Another intermarket that has a major effect on the currencies is T-Bonds. In the December 1993 issue of Formula Research, Nelson Free-burg discussed the link between T-Bonds and the currencies. T-Bonds and

TABLE 1.1

T-BONDS VERSUS

VARIOUS

INTERMARKETS.

Stock Group

Relationship to T-Bonds

S&P500 Chemical Croup

Negative

S&P500 Aluminum Index

Negative

S&P500 croup Steel

Negative

S&P500 Oil Composite

Negative

S&P500 Saving and Loans

Positive

S&P500 Life Insurance

Positive

foreign currencies are positively correlated. On a longer-term basis, this relationship makes sense. When interest rates drop, Dollar-based assets become less attractive to investors and speculators. Foreign currencies gain a competitive edge, and the Dollar begins to weaken. Freeburgs research has also shown that the link between T-Bonds and foreign currencies is


FIGURE 1.17 The Dollar index, Comex gold, and the CRB index weekly from mid-1992 to early 1995. The breakout in the CRB and gold in early 1995 was linked to a double top and then a collapse in the Dollar.



1 2 3 4 ... 16
© 2003 GARUN.RU.
Копирование материалов запрещено.