Trading in the stock, foreign exchange and futures markets, stock market, futures market, foreign exchange market, foreign exchange fututes

This book is about how to improve your trading in the stock, foreign exchange and futures markets. Although many technical solutions and spreadsheet applications appear in these pages, Smarter Trading is really about making decisions and solving problems. It will identify why many trading strategies and forecasts fail and will show how to improve results and create more lasting solutions.

The approach taken here tries to be realistic; trading systems have limitations, as do the tools and the traders. The techniques for improv ing profits and assessing risk focus on those areas that offer the greatest improvement, rather than the subtleties of fine tuning. Most of the more difficult topics are concerned with risk. Experienced traders usually know what to do with profits; even for novice traders, profits often take care of themselves. It is an unreasonably optimistic attitude toward risk that gets many traders into trouble; therefore, sections of this book keep returning to risk evaluation and control. We're in this for the long term.

Recent years have seen political and economic changes of large proportion. The emergence of China , the instability of the European Monetary System, and the faltering of Russia are all poised to produce massive changes in trade. At the same time, technology has made immense advances. More powerful computers come in smaller packages. Prices can be displayed for any time period in an array of multicolored windows.

Methods that once worked do not work anymore. IBM stumbles, pat terns change, markets are more volatile than ever, and even seasonals are not the same. Program trading has been declared "disruptive" all over the world. Traders sit behind screens in massive bank trading rooms, surrounded by high-powered displays, looking for arbitrage opportunities between any two markets in any two countries. All this continues 24 hours each day.

This evolution of markets is a structural change that moves in only one direction. The introduction of the automated exchanges in the United States and United Kingdom aren't anomalies, but a trend. Eventually, the floor traders will disappear—not all at one time, but edged out by automated exchanges that will surround them and slowly infringe on the sanctity of even the largest trading floors. Adapting may take more effort than simply retesting a program, adjusting for infla tion, or changing the value of a stop-loss, but it cannot be ignored.

Along with increasing complexity is additional competition. Floor traders once had the advantage of being aware of every price tick. Now we can recall prices and volume instantly, and catch up on market action even when we have been away from the picture for hours or days. It requires more to compete successfully. This book will help identify the problems. It will provide solutions and an understanding of new tools and how to use them.

Advances in technology have caused great changes in the trading industry. New tools and techniques are absorbed quickly. The goal of improving returns by a fraction of a point has tremendous rewards,

enough to motivate and finance major research projects. And as machines increase in their ability to process more of everything, push ing those limits becomes a compulsion.

There is also a fascination with the new tools and the ability to dis play graphics and analyze prices. Real-time data from all over the world is being processed at speeds measured in nanoseconds. We can do more faster and cheaper, even when we're not always sure what we are doing.

"Fundamentals" will always be the reason for price change. Fundamental analysis is the study of information that can influence cor porate earnings, dividends, and interest rates, resulting in a price change. In both stock and commodity markets, forecasts are the result of comparing current and past economic data, and determining the effects of government policy on interest rates and growth. Unexpected events introduce volatility and uncertainty.

In general, the direction of stock prices is related to the health of busi ness, which might include data on the Gross National Product (GNP), Consumer Price Index (CPI), retail sales, employment, and interest rates. If the economy is expanding, you can expect equity markets to rise. It is still a challenge to look further for the industries and sectors that will perform independently, or without correlation, to the market as a whole.

Supply and demand determine the price of goods and materials: how much there is versus how much is wanted. The more material, the lower the price; the more that is wanted, the higher the price. The fundamen tals of price change are clear. It is the changes in those factors or the anticipation of change that causes prices to move.

One of the first computer applications for price forecasting used a technique called multiple regression ( Box 1-1 ). Data on imports, exports, production, consumption, interest rates, inflation, technology, and other essentials could be analyzed in conjunction with the price. Fundamental analysis explains what has happened in the past by assign ing "weighting factors" to the information put into the computer. If you don't put in the right data, you don't get a good answer. Putting in too much data isn't as bad, but it takes a longer time to process. Sometimes, too much data allows the computer to find answers that are only coin cidental, which we call "overfit."

The answer to a regression analysis is assigned a confidence level, which indicates its accuracy. It is stated as "plus or minus" an error fac tor (e.g., interest rates will drop to 5V 2 percent ± 1 percent by May).

A simple regression model calculates a commodity price from the his tory of fundamental factors. In this example, the price of soybeans is simply the weighting of supply and demand:

where Est_Price is the current estimated value supply is the total production demand is the total distribution error is the error reflecting the accuracy of the results and constant, weights, and weightD are calculated using a regression program.

Data from 1964 to 1975 gives constant = -1.64, weights = 3.97 and weightD = 0.81, indicating that changes in the supply of soybeans, rep resented by the much larger value of weights, are much more impor tant than demand. The error factor would be large because only a few years were used in the calculation.

A regression analysis of fundamental data tells where prices should be, in the same sense as an option fair value calculation. It is very struc tured and provides only a single price range, which is considered "nor mal." It may not include data that cause the market to anticipate changes.

In general, this classic approach is not helpful to a trader because it says nothing about risk. If the current price level is below the calculat ed one, but prices start to fall instead of rise, when do you say that something is wrong? The forecast only shows where the price should be; it does not tell anything about how it will get there.

Fundamental analysis still makes sense, but it remains the domain of institutions and long-term traders. It requires a well-capitalized investor to absorb fairly large equity swings during periods when less important factors cause market volatility. For others, the risk is too high, the analysis takes too much time and effort, and the profits are too far off.

This method of econometric analysis can be applied to the stock index, but not to individual shares; however, results are still very general.

Newer methods, such as neural networks and expert systems, which are discussed in later chapters, have replaced econometric analysis, offering some additional accuracy when used in the traditional way, and much more flexibility.

Technical analysis is a broad area that uses price and related data to decide when to buy and sell. It tries to bridge the problems that funda mental analysis has about the specifics of timing and risk. The methods used can be as interpretive as chart patterns and astrology, or as specif ic as mathematical formulas and spectral analysis.

From the time chart interpretation first appeared to the early use of the home computer, technicians have been separated from fundamen talists. More recently, the line between them has become gray. Fundamentals provide the reason and direction of price movement, and technicals give the timing and risk control.

The two methods can be kept separate by making a decision and com mitment based on fundamentals, then creating and implementing the plan using technicals. Or, the process can be integrated with an expert system approach, a sophisticated neural network, or simply a combina tion of individual programs. It is only necessary that each make sense and satisfy the preset objectives.

Computerized technical analysis is associated most with the moving average ( Box 1 - 2). It came into greatest popularity in the early 1980s and has remained the basis for many technical programs. All factors that influence the market are assumed to be netted out as .the current price. A simple moving average applied to those prices gives a trend. The longer term 200-day average is a benchmark indicator of price direction for most stock issues. Short-term 5-, 10-, and 20-day trends, related to weekly and monthly periods, are often used for timing entries and exits and for leveraged futures and options markets.

A moving average, or similar technical indicator, is frequently used to confirm a decision to enter the market. Although there may be reason to believe prices will move higher, fundamental analysis can be com plex, and occasionally unexpected external factors overwhelm the normal situation. By waiting for a moving average to turn up, the program may sacrifice some initial profit for a greater chance of being correct and of using capital effectively. A moving average is now a "function" in a strategy testing program or a spreadsheet. A 3-day moving average may appear as:

@Average(close,3) = (close + close[1] + close[2])/3 or the spreadsheet form: @AVG(B3..B5)/3

An exponential trend is simply another function, @Exp_MA(close,10), all of which makes price analysis very easy to perform. The trendline formed by the daily averages produces a buy signal when it turns up and a sell when it turns down.

Trend trading was very successful during the 1970s and into the 1980s. Even now, it is just as important to know the direction of prices, but it has become more difficult to trade a simple trend system. What was once a strong commitment to "the trend is your friend," we now hear that there are too many systematic trend-followers who are "push ing the market" and "triggering stops."

Trend following proved to many investors that technical analysis is viable. Some analysts, equipped with real-time intraday data, were able to continue trading the trend by applying the same logic to hourly or 15-minute prices. Others looked for additional indicators or more sophisticated tools, often adapted from another industry.

Timing indicators, such as relative strength and stochastics have become very popular. Most quote equipment provides a broad selection of techniques that users can modify. The Relative Strength Index (RSI), devel oped by Welles Wilder, is a ratio of the total daily upmoves to total daily downmoves over the past 14 days, expressed from 0 to 100:

@RSI(close,14) = 100 x (RS/(1 + RS) where RS = @SUM(TotaLUp_Moves,14)/@SUM(Total_Down_Moves,14)

Because of the flexibility of computers, any trader can substitute anoth er time period for the standard 14-day interval.

The stochastic, developed by George Lane , is equally popular. It gives the relative position of the closing price within the previous high-low range, determined by the length of the period used for the calculation. The raw stochastic, called FastK, which ranges from a value of 0 to 100, is not normally used for trading because it is too sensitive to price change. Instead, the SlowK (also called %D) and SlowD, a 3-day smooth ing of SlowK,- have become the popular values for trading. The 5-period stochastic is given as

@SlowK(series,5) = @Moving_Average(@FastK(series,5),3) @SlowD(series,5) = @Moving_Average(@SlowK(series,5),3)

As with the RSI, traders vary the period of the stochastic to make it more or less sensitive to price movement.

In addition to indicators that use price, a wide selection of calculations are based on volume, or the number of advancing and declining stocks. It is interesting to see how the same numbers are used with slightly different emphasis:

For each of the indicators, an index is created by adding the current value to the accumulated index value for the previous day, as in the Bolton-Tremblay Index:

While there is virtue in simplicity, it is not clear that an index such as the one based on advancing and declining issues improves a moving average signal. The indexes themselves require interpretation and selection because they do nothing to filter noise, nor is it apparent how to use those days with significantly greater advancing or declining issues. Instead of just one difficult price series to work with, you now have the original price data plus an equally difficult index.

Simple moving averages and indicators do not give the complete picture of technical analysis. There are sophisticated graphics programs that allow the user to draw trendlines and channels. They can perform spectral analysis to find cycles; give Gann lines and angles, Fibonacci spirals, and Elliott waves; and create ARIMA (Autoregressive Integrated Moving Average) models, which constantly recalculate the best moving average for every new piece of data.

Testing software has also pushed the industry forward. Anyone with an untried trading strategy or theory can enter simple instructions and test the rules on a wide selection of data. The most popular examples of this software are Dow Jones' TeleTrac, Omega's System Writer, and Equis' MetaStock. These programs eliminate the need for a computer specialist, by providing standard "functions" (as described earlier in this chapter) for calculating a moving average, true range, highest and lowest price, and many other convenient values, in an easy-to-use form. For example,

@Average(Price,Length) @Bollinger_Band(Price,Length,StdDev) @Bullish_Divergence(Price,Osc,Strength,Length) @Linear_Regression(Price,Length)

are the functions for a simple moving average, a Bollinger band, bullish divergence, and the angle from the horizontal of a linear regression line. They are all easy to understand even without a manual. Length is the number of periods used in the calculation, and price is any data series, whether daily, hourly, or user defined.

Strategy-testing software takes two forms: (1) a spreadsheet style sys tem, such as TeleTrac, in which each calculation becomes a new row, and each day appears as a column; (2) the System Writer design, which allows all the power of computer programming, similar to the BASIC language (Box 1-3). In Omega's software, the individual calculation can be dis played by day on request.

An "efficiency ratio" is calculated as 10-day price changes divided by the sum of 10 daily closing price differences. The EfRatio gives the effi ciency, or relative noise, of price movement over the 10-day period, and is used for the Adaptive Moving Average in Chapter 8.

TeleTrac allows the calculations to be entered in their own spreadsheet format. The first column is the name of the value being created and the second gives the formula. Results are shown for each time interval selected, from ticks to monthly. This example has daily data:
18Feb93 23Feb93 24Feb93 25Feb93

System Writer uses Omega's fully programmable Easy Language to find the same ratio:

noise = @Summation(@AbsValue(close[0] - close[1], length) signal = close[0] - close[10]

The last line of the System Writer code tests that the noise is not equal to ("<>") zero before dividing. In the first line, it "nests" the first two lines of the TeleTrac code.

Buy and sell signals, as well as moving average lines, oscillators, and user-defined values can be displayed on split-screen graphics. Every calculation and detailed statistics of the test results can be easily retrieved and printed.

As computers have become more powerful and scientists try to synthe size the human brain, some new techniques have become more practical for market analysis. The area called artificial intelligence includes the most promising neural networks, as well as simple pattern recognition, expert systems, and fuzzy logic. These will be covered later in this book.

Neural networks are already replacing regression analysis as the best method for finding how fundamental factors change price. But the power of neural nets is still untouched. It is primarily used to find the same continuous relationships between dividends, economic condi tions, and price—or supply, demand, and price—that was the tech nique of the older tools. It has the ability to identify how today's set of factors compares with specific cases in history, an equally sophisticated and practical approach.

It is easy to argue that communications and computers have changed world and local economies, trading tools, and market participants. The extent of these changes requires us to look at the process as an evolu tion. It should not be a surprise that simple systems that worked well during the 1970s and 1980s are no longer profitable.

Economic changes have altered price relationships. Countries have floated their currencies and interest rates, as gold was allowed to be traded in the United States in 1975. Other countries have tried to control the fluctuation by forming pacts, such as the European Monetary System or OPEC (Organization of Petroleum Exporting Countries). For each of these events, there is a change in price patterns, sometimes more volatile; other times less volatile; and still other cases go unnoticed. A simple way of looking at this evolution is by comparing the perfor mance of a trend-following system in different types of markets.

As markets become familiar and accepted, participation increases, and with more participation comes a higher level of "noise." The U.S. stock market has been a main investment for pension funds, individuals, and other countries for many years. Broad interest takes on the characteristic of a constant level of activity resulting from unrelated objectives of its participants, much like the undercurrent of constant talking at larger and smaller meetings. A pension fund adds to its positions because of new investment capital; a corporation liquidates a stock portfolio to invest in a shopping center, or a family withdraws from a mutual fund to pay medical bills. Much of the price movement has little to do with economic trends or clever timing of entry and exit points.

New or emerging markets do not have this broad participation; there fore, the level of noise is not nearly as high. That makes it easier to trade a simple trend-following system profitably. Table 1-1 and Box 1 - 4 com pare more and less mature markets. They show that it is easier to trade a newer market with a simple system, and just as impossible to use that same approach for a broadly traded, mature market.

Results show that the newest, active world markets, the Hong Kong Hang Seng Index and MATIF's CAC-40 are still very profitable for all trend speeds tested. The Hang Seng has a higher annualized return for faster trends, indicating that short-term price swings are sustained with a relatively low level of noise, even though the longer term trend did not develop as well. Had the noise been greater, the trend would have reversed frequently, causing whipsaw losses.

The CAC-40 shows a drop of about 50 percent in trend returns for the most recent five years compared with the earlier period. It also posted larger profits in faster trading for the earlier period but has evolved to a more uniform distribution as the market matured. There is now only a slightly declining pattern in the rate of return as you move from the shortest 5-day trend speed to the longest 75-day period.

The Deutsche mark, along with other major currencies, has always been actively traded. In the past five years, however, Deutsche mark activity has increased as the mark assumed a dominant role in the European economic structure, combined with greater world trading activity. The first five years, 1983-1987, show greater profits and a tendency toward longer term trends than the more recent years. Both peri ods show that the fastest trends are the most difficult, the likely result of short-term volatility, frequent price shocks, and intense competition in active Forex trading. The "optimum" trend may vary based on long-term economic shifts, confidence in the European Community, and the ability of the United States to compete in trade. As the foreign exchange markets continue to mature, the ability to produce trending profits with faster trading should decline.

IBM is shown for only the first five-year period, to allow you to study its change. With the exception of the 5-day trend, it shows a clear increase in profits as a longer-term position is taken. With transaction costs included, the smaller profits posted in the 15- to 45-day trends are likely to disap pear. The profit of 18 percent for the 5-day trend may seem unusual but would have presented a window of opportunity for some small investors.

Changes in price patterns as markets mature can be illustrated with a simple trend-following system. It shows that newer or emerging mar kets have clear trends. As they mature, the increased participation increases noise and makes the price patterns more complex. Orders entering continually, for different reasons, obscure the trend.

Table 1-1 shows the results of a simple trend-following system tested for a selection of markets. The system uses the following rules:

	Hang								Crude
Exp	Seng	CAC-40		D-Mark		IBM		Dow	Oil
MA	1988-	1983-	1988-	1983-	1988-	1982-	1983-	1988-	Dec
Days	1993	1987	1992	1987	1992	1987	1987	1992	1992
5	22.3	40.5	16.6	5.7	1.3	18.0	(47.5)	(36.2)	(0.17)
15	17.5	33.1	20.7	9.7	4.9	4.0	(22.3)	(26.8)'	0.27
25	12.1	32.8	19.9	6.7	7.4	4.3	(22.7)	(16.1)	(0.41)
35	9.8	30.0	19.7	9.3	6.2	6.9	(18.9)	(14.9)	(0.43)
45	8.7	29.1	10.8	8.5	9.3	8.6	(12.9)	(14.4)	(0.91)
55	11.7	28.5	13.5	9.3	6.3	11.7	(0.7)	(8.7)	(1.19)
65	11.1	27.7	18.3	11.7	5.8	11.1	0.8	(6.6)	(0.85)
75	18.0	27.8	13.5	11.2	5.8	18.0	(0.7)	(11.6)	(0.17)
Avg	10.4	31.2	16.6	9.0	6.2	10.3	(15.6)	(16.9)	(0.48)

• The most mature market, the Dow, cannot be traded using any
trend within the tested range.

• Developing markets, such as the Hang Seng and CAC-40 (from
1983 to 1987) show characteristic short-term trends. As they
mature, they begin to show greater profits in long-term trends,
rather than short-term.

• The D-mark is a deep market, as are all the other major curren
cies, and shows a tendency toward declining, smaller profits.

• IBM shows long-term trends with the exception of a small oppor
tunity window for the 5-day trend, which is difficult to trade.

• Crude oil volatility, the result of large commercial dominance
and attempted OPEC manipulation, overwhelms the price trend.

We might assume that institutions cannot trade a short-term trend because they cannot trade a large enough position and tend to move the market too far for the expected profit. Small, independent traders can take advantage of this slot only if their commissions and other transaction costs are low. A 5-day trend may have traded as much as once each week. With commissions at V 2 percent, an 18 percent profit becomes an 8 percent loss for the year.

The Dow represents the most mature market. The large losses at near ly every trend speed indicates that the daily market volatility is very high compared with the net movement of stock prices. Mature markets have broad participation, representing traders with diverse interests. Constant institutional buying and selling to fill portfolios, or specula tive liquidation to invest in real estate or other assets, adds a high level of noise to the daily activity. As other world markets mature, they should follow the pattern of the Dow.

The last example looks at a single futures contract of NYMEX crude oil. It also shows losses in nearly every trend period. But crude oil is not as much a mature market as it is a manipulated one. Attempts by OPEC to limit production cause sharp moves and discourage small, individual traders from participating. The cash and futures market, dominated by very large commercials, are used to buy and sell cargoes for delivery and hedging. This combination—attempted price controls and traders with large orders and deep pockets—adds volatility that overwhelms the trend.

Can you make the same type of trading decisions today as you could in 1960? No. Many of the markets with the greatest volume form a relatively new "derivatives" group. Foreign exchange futures, options on futures, stock index futures of all types in many countries, and short- and long- term interest rates, including the European Currency Unit or ECU (which is more a concept than a hard currency that you could use to pay a lunch bill) are all very different from the markets only 10 years ago.

Even agricultural markets, the oldest of the exchange-traded com modities, have changed. Crops are still planted and harvested accord ing to their season, but the seasonal price patterns that survived for cen turies are no longer dependable. In the 1980s, U.S. grain prices rose with surplus production, confusing traders and delighting farmers. Afterward, the phenomenon was recognized as "parity," the ability of a freely traded product to hold a constant world value. As the U.S. dollar declined, U.S. grain became more attractive to non-U.S. buyers, causing export demand and price to rise.

It seems odd to think that the normal seasonal patterns have also changed. Beginning in the mid-1970s, fanners built more storage facili ties. Before there was adequate storage, farmers needed to forward price their production with a local grain elevator or be subject to selling everything that couldn't be stored at harvest. They could rent storage space, but only if they booked in advance and paid a minimum 3-month fee. On-farm storage relieved the pressure on harvest prices and farm ers could recoup their cost by adding carrying charges to the crop price each month after harvest. With 100 percent storage, it is possible to eliminate the price drop at harvest.

With their own storage facilities, carrying charges become "soft dol lars" to the farmers. When crops were stored in a rented space, fanners paid out-of-pocket costs to elevator operators. On-farm storage is simi lar to price "bundling"; it is harder to distinguish the cost of the parts, and the total price becomes more flexible.

Orange juice presents two more seasonal wrinkles. Crop freezes are no longer devastating to consumers. Prices may jump sharply when temper atures hold under freezing in central Florida , but not for long. For the past few years, Brazil has been anxious to fill the gap in U.S. supply, and consumers are equally agreeable to paying lower prices at the expense of domestic U.S. growers. Because the government bowed to the pressure and permitted juice to be imported from tremendous Brazilian reserves, what will happen the next time there is a freeze? Do prices jump on a sharp reduction in supply, or will traders anticipate a fast substitution of Brazilian product? In either case, the seasonal pattern is disrupted.

The second twist is the result of the opposite North American-South American crop years. Orange juice, soybeans, and eventually other crops can be produced in quantity in the opposite season. When the U.S. farmers are harvesting in the fall, Brazilian farmers are planting in the spring. Just when carryover stores are dwindling, crops from the other hemisphere are being harvested. When U.S. crop production is not good, Southern Hemisphere stocks and plantings can be abundant. The world import mar ket is fungible. If the Russians need grain, they will go to the best price, not just to the U.S. market. What does this do to the seasonal patterns? It makes supply and demand a global concept, and far less predictable.

This is not just an agricultural phenomenon. It is no secret that financial markets have been globalized. These effects are permanent. Even if gov ernments change and exchanges disappear, the United States will never again be the only center of activity. Undoubtedly, the People's Republic of China (PRO and Russia will play an increasing role in world trade. Despite the PRC's apparent resistance to political change, it has shown an aversion to cutting off its Hong Kong arm. Regardless of inconsis tency, China —with a quarter of the world population—wants to enter into world trade more actively. Russia has already made it clear that it wants a market economy. The effects on supply-and-demand equilibrium will be enormous. If the transition is fast, the imbalance will provide short-term opportunities in business and speculation that are unprecedented. Either way, the future will not look the same as the past.