Forecasted Moving Averages:
Creating Leading Indicators Through Intermarket Analysis

When you review all of the technical indicators available to traders in today's analytical software, moving averages are still one of the most popular and widely-used indicators to help identify market trends. Moving averages form the basis of many single-market, trend-following trading strategies, ranging from the popular 4-9-18-day moving average "crossover" approach to the widely followed 50-day and 200-day simple moving averages used to highlight the underlying trend direction of broad market indexes and individual stocks.

Moving averages, calculated according to precise mathematical formulae, are an objective (quantitative) way to ascertain the current trend direction of a market and develop expectations about its future direction. Moving averages filter out the random "noise" in past price data by "smoothing" or "averaging" out the fluctuations in price movement.

Lagging Behind
However, traditional moving averages have one very serious deficiency: They are a '"lagging" technical indicator. By virtue of their mathematical construction (averaging prices over a number of prior periods), they have to rely on past prices that have already occurred so tend to lag behind the current market price.

"Making trades based upon the analysis of moving averages typically results in getting into and out of the market late when you compare the points at which the market's price actually makes a top or bottom and when it changes trend direction," points out Lou Mendelsohn, developer of VantagePoint Intermarket Analysis software. "Depending on the market's price movement and the type and size of moving average used, this lag effect can be substantial, causing the difference between trading success and failure in today's highly volatile, global financial markets."

Notice, for example, how the moving average lags behind the market at major turning points on Figure 1, a chart of daily prices of the U.S. Dollar Index futures contract with its actual 10-day simple moving average.

This lag is the Achilles' heel of moving averages. Technical analysts have spent years on research in a futile effort to eliminate this lag while still retaining the beneficial "smoothing" effects of moving averages. As a result, numerous types of moving averages have been devised, each with its own mathematical construction and effectiveness at discerning the underlying market trend and ability to minimize the lag effect.

Complexity Doesn't Help
Moving averages are often incorporated into more complex technical indicators, such as moving average crossover strategies, to improve their effectiveness. One such approach involves two simple moving averages of different lengths, such as a 5-day and a 10-day average. When the short moving average value is greater than the long moving average value, the underlying trend is assumed to be up. When the short moving average value is less than the long moving average value, the trend is assumed to be down.

The chart of the New York light crude oil futures contract provides an example of how the shorter 5-day moving average is more responsive to current price action than the longer 10-day simple moving average (see Figure 2), but both still lag behind the market at major turning points.

An inherent assumption behind moving averages is that once a trend is underway, it tends to persist. Therefore, until the long moving average is penetrated by the short moving average in the direction opposite from the prevailing trend, an uptrend is assumed to remain intact.

Traditional moving average crossover strategies are effective at identifying the current market direction in strongly trending markets. In non-trending, sideways markets, however, and even in trending markets when very short moving averages may be overly sensitive to abrupt price fluctuations, these approaches are subject to whipsaws. This results in erroneous trading signals at market tops and bottoms. So, while traders can make money in trending markets using moving averages, it is the choppy markets, increasingly more common today, that can cause substantial trading losses.

Various crossover strategies have been created in a further effort to overcome this deficiency. One popular approach compares an actual price, such as the daily close, with a moving average value. Other commonly-used approaches attempt to minimize whipsaws and filter out faulty signals by using bands surrounding the moving averages, using three or more moving averages,! or combining moving averages with other single-market technical indicators for additional confirmation.

With today's unprecedented intraday and interday market volatility, caused in no small measure by the globalization of the markets and resulting effects of related markets on one another, traders can no longer rely solely on single-market lagging indicators such as moving averages. Knowing that a market made a top or bottom several days ago is no longer an effective way to make trading decisions, if it ever was. Even a one-day lag in today's fast-paced, globally interconnected markets is too long to wait for this information.

"It is imperative that traders adopt an intermarket perspective and incorporate intermarket data into their current trading strategies, so they can develop effective leading indicators that correspond to how today's global financial markets really exist," Mendelsohn contends.

New Way to Forecast
The purpose of technical analysis is to identify the underlying market trend and forecast (or at the very least extrapolate) its future course for the purpose of making profitable trading decisions. Therefore, it seems logical that this goal could best be achieved through applying leading indicators that use both single-market and intermarket data, rather than by continuing to rely upon trend-following indicators such as traditional moving averages that are computed solely on past single-market data.

Theoretically, a predicted moving average value for a future date, if it were 100% accurate, would have, by definition, no lag whatsoever. Since this is impossible, something else must be done to bring this widely-used trend-following indicator into the 21st century of trend forecasting.

One innovative solution to this dilemma transforms moving averages into a leading indicator by using both single-market and intermarket price, volume and open interest futures data as inputs into the design of neural networks, which are then trained to make short-term forecasts of moving averages.

"Neural networks are a mathematical technology from the field of artificial intelligence and can be trained to find reoccurring patterns and relationships within both single-market and intermarket data that can be applied to market forecasting," Mendelsohn explains.

"These forecasted moving averages are then incorporated into predictive moving average crossover strategies that identify market trend direction of individual financial markets with very high accuracy."

For instance, to forecast the short-term trend direction of 30-year U.S. Treasury bond futures for the next several days, neural networks can be trained on past single-market data on the 30-year U.S. Treasury bond itself, in addition to intermarket data from various related markets. Sophisticated software can be used to analyze related markets to forecast several moving averages of different lengths and of different forecast time horizons on 30-year Treasury bond futures. This could include a 5-day average for two days in the future and a 10-day average for four days in the future. The related markets would include the 10-year U.S. Treasury notes, U.S. Dollar Index, Euro FX, Comex Gold Index, S&P 500 Index, Japanese Yen, Eurodollar, Bridge/CRB Index and New York light crude oil.

Figure 3 shows a crossover of a predicted 10-day simple moving average for four days in the future with today's actual 10-day moving average for 30-year T-bond futures. Notice that the predicted moving average, because it is forecasted in advance, does not lag behind the market, while the actual 10-day average lags behind both the market and the predicted average.

Each day as the neural networks are updated with the most recent single-market and intermarket data, new moving average forecasts are made, and the difference in value between each predicted and actual moving average of the same length is determined.

Crossover Signals That Pay Off
In this simple example, when the predicted moving average crosses the actual moving average from above to below (the difference goes from positive to negative), the market trend is expected to turn down within the forecast time horizon. When the difference reaches a maximum negative value and starts to narrow (indicating that the downward trend is beginning to lose strength), this is an early warning that the market is poised to make a bottom and turn up.

Rather than wait for the crossover itself to occur, trading decisions can be based on a narrowing in the difference. For instance, when the difference reaches a maximum negative value and starts to narrow, you can act on this information in a number of ways depending on your account size, risk propensity, trading style and objectives. Mendelsohn cites just three possible scenarios that can be pursued (assuming that you are in a short position):

• If the difference reaches a maximum negative value and narrows by even a small amount, you can close out your short position and stand aside. Then you can wait for the difference to narrow further before going long.

• If the difference reaches a maximum negative value and narrows by even a small amount, you can tighten your stop and stay in your short position until the next day.

• If the difference reaches a maximum negative value and narrows by a minimal amount, you can close out your short position and go long. This strategy is the most aggressive of the three because it involves reverising positions at the earliest indication that the current market trend is expected to make a bottom and change direction.

No Financial Crystal Ball
By employing leading indicators, using both single-market and intermarket data such as forecasted moving averages, early warnings of imminent changes in trend direction become apparent days before they show up on traditional price charts or can be identified by single-market trend-following indicators that lag the market.

Admittedly, it is impossible to create leading trend forecasting indicators that can forecast future market direction with 100% accuracy. This elusive Holy Grail is the financial market equivalent of a desert mirage. In reality, no more than 80%-85% predictive accuracy can ever be achieved, given the randomness and unpredictable events that are inherent in today's globally interdependent financial markets, as well as due to the daunting task of creating effective forecasting tools that can stay current with rapidly evolving, complex financial markets.

As the futures and equities markets become even more intertwined and more traders incorporate intermarket analysis into their trading strategies, powerful leading indicators, such as the predictive moving average crossover strategies that expand upon the concepts of popular trend-following indicators, are a must for serious traders. These indicators will allow traders to seize trading opportunities based on predictive information derived from the hidden relationships and complex patterns among related global markets.

About Today's Author
Darrell Jobman is an acknowledged authority on the financial markets and has been writing about them for over 35 years. After spending nearly 20 years as editor of Futures Magazine Mr. Jobman is now Senior Market Analyst for TradingEducation.com. Mr. Jobman has authored and/or edited six books including The Handbook of Technical Analysis as well as trading courses for both the Chicago Mercantile Exchange and the Chicago Board of Trade.