Tips on Technicals - MACD
Technical analysts often apply studies to the results of other studies; moving averages of spreads and RSI's of RSI's, to name a few. Technician Gerald Appel took a simple departure (oscillator) chart, which measures the distance between two moving averages, and overlaid a moving average of the result. This study was named Moving Average Convergence-Divergence (MACD) as it measures how two moving averages (the departure chart) move together and apart over time.
The underlying theory is that as a market moves higher, the shorter of two moving averages is above the longer average. This is because the shorter average reacts faster to price movements. The longer average will always lag behind. When the shorter average crosses the longer, it is a signal that the trend may be reversing. Viewed in a departure chart, this crossover occurs when the two averages have the same value and therefore touch the zero line. Figure 1 shows 200 days of daily bars for the Dollar-Yen spot rate with 26 and 12 day moving averages. As the market falls, the 12 day average is below the 26 day average. When the 12 day crosses above the 26 day, the departure study below crosses through zero. The departure study shows the difference between the two averages on the chart.
Figure 1
MACD starts with the departure chart and figuratively overlays a moving average of the departure values. Notice how the MACD line (solid line) has exactly the same shape as the histogram of the departure study. The dotted line is called the Signal line and is just a moving average of the MACD line.
Using MACD
The typical MACD analysis uses exponential moving averages. Like other oscillators, MACD works best in a trending market or one that is trading in a volatile trading range. In flat and quiet markets MACD can produce unreliable results and often disagrees with basic trendline analysis.
The first indication of a change in trend is divergence. This is when the market makes higher highs and the MACD makes lower highs. The converse is true for falling markets where price makes lower lows and MACD makes higher lows. (See the complete discussion on Divergence.) Figure 2 shows one year of daily data for NYCSC December Cocoa futures. The divergence during the summer of 1994 is clear and the market broke its uptrend line by Autumn.
Figure 2
The second indication can be either a crossover of the MACD and Signal lines, which is good for trending markets, or a trend break in MACD, which is good for volatile markets. The Cocoa market mentioned above was relatively volatile during its divergent period. Buying and selling points were indicated throughout the chart when the Signal line crossed above and below the MACD line. Just like with moving averages and departure charts, this crossover indicates a possible entry point. Note that crossovers should occur at overbought or oversold levels of MACD. Since MACD is not indexed (like RSI or Stochastics), the definitions of overbought and oversold are market specific. For this discussion, lets just say that when MACD is at a relatively high level and the signal line crosses below it, a sell signal has been generated.
The volatile market MACD method is analyzing trend breaks in the MACD itself. Figure 3 shows one year of daily data for Knight-Ridder stock. Each trend break in MACD preceded the trend break in prices by several days. As can be seen in the July 1994 break, MACD gave a short term signal that the bar chart did not. In this case, MACD failed to remain above the zero line as it met the stronger resistance of the down trend line in prices.
When both MACD and Signal lines cross the zero line, it is a signal that the trend reversal is complete. A market that shows divergence first, crossover or trend break second and zero line crossover third has given a very strong indication that a new trend is in force.
The MACD Histogram
Further analysis can be made by charting the difference between the MACD and Signal lines (not shown). This is the same as plotting the departure (oscillator) of the two lines and it is typically drawn in histogram format. This makes it easy to detect extreme ranges of departure as well as when the two lines cross.
MACD is similar to other studies but adds another dimension to analysis. It should not be used alone but rather in addition to trendline analysis.
