Trading Basics—The Directional Movement Oscillator
11/20/2008 9:43 am EST
In earlier articles, I demonstrated how the average directional movement index (ADX), developed by Welles Wilder, could be used to identify trending and non-trending markets. Though the ADX formula is fairly complex, it is derived from the DI+ and the DI-, which can also be used to generate buy and sell signals.
As is the case for each of the indicator's featured here, I like to use them on multiple time periods for the same market. This is a technique used by many traders since just taking the signals based only on the time period that you trade can result in too many trades against the dominant trend, often resulting in losses. In the chart above, we have a weekly chart of Boeing (BA) from early 2000 through the first months of 2002. Below the bar chart, there are two lines: the nine-week minus DI (or DI-) in red, and the nine-week DI plus (or DI+) in green. I prefer to use the nine-period on the weekly data and the fourteen-period with the daily data. The most basic interpretation is that when the DI+ is above the DI-, it is positive; when the DI+ is below the DI-, it is negative. When both lines are close together, and particularly when they are declining, it is consistent with a non-trending market. In the latter part of May 2000, the DI+ moved above the DI- at point 1. The two lines stayed well apart until early December as the DI+ started to decline and crossed below the DI- on the first week of 2001. For the next three months, both lines were flat and below 20 until the week of March 16, 2001 when the DI- moved above the DI+. Though BA did decline the next week it turned higher the following week, and three weeks later the DI+ crossed back above the DI- (point 3). This positive signal lasted until the middle of June when the DI- rose sharply, surpassing the DI+ at point 4. During the next few months, the DI- rose very sharply consistent with a trending market. This negative signal stayed in effect until January 2002.
It is often easier to visualize the crossover points by viewing the DIosc as a histogram, so I have added the DIosc in histogram form to the bottom of the chart in Figure 2. You will also note that short-lived signals, such as those given in early 2001, are more easily identified. However, I recommend that you plot both the lines and histogram, as it is important to note the slope of both lines, as well as their actual values. Neither of these qualities are available in the histogram version. Another thing that I have changed on this chart is that I have used candlesticks instead of a bar chart. I have noticed that significant candle formations often accompany the crossovers in the DI+ and DI-, and for this reason, I have decided to include candlesticks. Because of the scaling, the bullish candle formation is not as easy to see on this chart (see point 1). So I have just highlighted the bearish formations as indicated by the circles labeled a, b, and c. If you examine these three periods, you will realize that candle reversal formations preceded or coincided with the DI- crossing above the DI+.
Welles Wilder advocates using what he called the "extreme point rule" with crossings of the DI+/DI. This adds a risk management component to the crossings; a critical ingredient to any trading method. The figure above has a weekly chart of Lehman Brothers Holdings (LEH) on the left and a daily chart on the right. Below each chart is plotted the DIosc, using 9 and 14 periods, respectively; because of space limitations, the plots of the individual DI+ and DI- lines are not shown. There are several periods on these charts that I would like you to examine in more detail, using the extreme point rule. The first occurred on June 10, 2005 (box a), as the weekly DIosc turned positive. For that week, LEH made a high of $47.89 and a low of $46.16, and the extreme point rule would, therefore, tell a trader that the positive signal would stay in place as long as the low of that week ($46.16) was not violated. This level is shown on the chart. Now, if you look at the corresponding box a on the daily chart, you will note that LEH corrected the first two days of the following week with a low on June 14 of $46.21. The strong close the next day provided strong evidence that the uptrend had resumed.
LEH continued higher for the next six weeks as it gained about 15% before closing a bit lower the week of July 29, 2005. This was the start of a period of consolidation as the daily chart shows a continuation pattern (lines 1 and 2). The weekly DIosc stayed solidly positive during this period while the daily DIosc dropped into negative territory several times as the triangle was forming. Since the weekly DIosc was positive only long signals would have been taken. Trading exclusively from the signals (ignoring the chart formation), the first daily buy signal came with the close on August 10. Long positions established the next day on the opening ($52.42) should have been protected with a stop under the August 10 low of $52.23. This stop would have been hit in the next few days as LEH dropped to a low of $51.55. Another positive daily signal was given on August 18 and longs established at $52.65; the following day's open should have used a stop under the 18th's lows of $52.10. LEH moved higher for a few days, reaching a high of $53.18 before reversing on August 24th and dropping to a low of $51.57. I am reviewing these trades in detail so as to illustrate that even though they resulted in losses, the extreme point rule kept the losses very small.
The next daily positive signal came on August 31, one day before LEH broke out of its triangle formation (lines 1 and 2). The next day's open was at $52.83 and the stop, once again, should have been below the 31st's low of $52.34. This rally lasted five weeks as LEH approached the $59 level. The lower close the week of October 7 (circle b) was accompanied by the daily DIosc turning negative. On Friday, October 14, the weekly DIosc also turned negative as the severity of the decline is evident on the daily chart (circle b). With both the daily and weekly DIosc negative, a short position should have been established at the October 17 opening of $54.45 with a stop above the high of the previous week, which was $56.34. This stop was hit two days later and the next week both the weekly and daily DIosc were again positive. The weekly DIosc stayed positive for the next seven months as LEH moved above $78.Fig 4
The same methodology can be used for those trading based on intra-day data. In Fig. 4, on the left we have a daily chart of the Yen/$ cash rate and to the right we have the 60-minute chart. Both have the 14-period DIosc plotted below them. The daily DIosc moved back above the zero line on January 9th, as indicated by line A. The low on the 9th was at 118.69, so this was the level that would have to be broken to reverse the signal. The hourly DIosc was at zero for the first few hours of trading on January 10th, but then moved above the zero line (line A) at approximately 118.21. The flag formation on the hourly chart also gave traders a good setup as the dollar broke out to the upside at 9:00 AM EST on the 10th. For most short term FX traders, a stop under 118.69 would have been too wide and they would probably have used one that was just under 119 or 118.90. Once the flag formation was completed, the stop could have easily been moved to under the lows of the continuation pattern. The dollar did not show any signs of losing upside momentum until January 23-24, as it closed lower both days.
I am sure many of you have already been using the DI+/DI-, but if you have not, I hope you will take the time to apply it to some of the markets that you follow. I have found it to work well with all markets and have even used it on the mutual funds in my 401K. It also works well with any time frame, although I would use nine periods for weekly data, and less periods for longer time frames; fourteen periods seem to work best for daily and intra-day data. In a future article I will show you some techniques for further refining the DIosc signals.
If you have a question about this article or would like to have a trading method discussed in more detail, I can be contacted at firstname.lastname@example.org.