I trade the daily time frame. I gave the Cowabunga a limited trial, figuring that if it worked on a lower time frame it should work even better on the daily, as there relatively less noise. Because as a trend trader I wouldn’tleave home without the ADX, I included it just for kicks. And in fact the Cowabunga worked quite well, as you can see from the chart—4 winners and one loser. However, I noticed a couple of things, one of which would have prevented the loser.
In the chart, green arrow denote entries, either Long or Short, and red arrows denote exits. The thick trend line shows the direction of the trade. Entries are mandated by the Cowbunga criteria of EMA 5 crossing 10, RSI above or below 50 as appropriate, and Stochastic rising or falling as appropriate.
First, the Stochastic seems to be superfluous. On each of the 5 trades it gave a go-ahead, even on the loser on 2007.10.15. If EMA 5 has crossed the 10 and the RSI agrees, the Stochastic will usually rubber-stamp the deal.
Not so the ADX.
On the Long entry, 2007.06.19, ADX has crossed upward above 20, the +DMI is dominant and the DMIs are diverging sharply. All is well.
Notice that I ran a Fibonacci Extension on the mid-trend retracement and set a Take Profit at FE 100, and the trend peaked there. I did this on all 4 winning trades and it always worked. If the 100 level had not been hit I would have exited on a turndown of ADX from a peak or on EMA 5 crossing back over 10.
On the Short entry, 2007.07.27, ADX has turned down from a relatively high value, and the -DMI has angled sharply up to meet the sharply falling +DMI. All is well.
On 2007.09.05 the Short entry taken because EMA 5 had crossed below 10, RSI is below 50, and Stochastic is falling. However, ADX is <20 and the DMIs have been whipsawing. All is not well. And indeed the trade turned in the wrong direction, hit the stop, and lost 316 pips. Note that the Stochastic, which is the method’s momentum oscillator, did not worn of this bad entry.
On occasion 2007.10.15, when Stochastic would have kept us from going Long on a 5/10 cross, so would the sub-20 ADX and whipsawing DMIs. The subsequent Long entry would have been confusing, since 5 had already crossed 10 without an entry, but ADX comes to the rescue. On 2007.10.29 ADX has swooped beneath both DMIs and risen 4 points (our usual criterion), +DMI has crossed above _DMI and they are angling apart. Stochastic, however remains pinned. It appears that for this application anything Stochastic can do, ADX can do better.
On 2007.10.15 a Short entry is mandated by an EMA 5\10 cross, an RSI slightly below 50, and a falling Stochastic, although the wimpy RSI position might give pause. We see that ADX has turned down from a high value and the rising -DMI is angling sharply toward +DMI but has not actually crossed it. ADX-wise we are technically still in Bear territory. This is a judgment call. Notice the Bear divergence shown on ADX and Price, less convincingly on RSI, and not at all on Stochastic. This means that a fairly large decline is probably in the offing and should tip the scale toward an immediate entry.
On the daily time frame at least, the method would seem to benefit from replacing Stochastic 10,3,3 with ADX/DMI 14. Incidentally, this would give us an all-Welles Wilder trading method.
But wait, there’s more. Because the RSI measures the averages of the up closes and down closes, it is a rather crude instrument, and this is evident on the chart. RSI often wavers almost to 30 and almost 70 when there is actually no trend, and so it is rather ambiguous when used on quick-changing mini-trends. If you automate it you will get a lot of false signals. Because the DMI and its derivative the ADX measure directional movement relative to true range, they are finer in their representation of trend and momentum. They rarely give a false reading. In truth, the RSI can also be dispensed with.
But this good news is also the bad news. The ADX/DMI contains a lot of information, but the elements must be integrated in order to realize their potential. Have the DMI’s been whipsawing? How far apart are they, and at what angle? How high is the ADX, and how steep? Are the peaks diminishing uniformly or raggedly? And so on. For this reason, it is not very feasible to automate the indicator in a high-confidence way. It must be read intuitively. And you must insert a Fibonacci Expansion every time there is a retracement, in order to get good exit points.
So where does that leave us? With Price, EMA 5 and 10, and ADX/DMI 13, which are to read manually. If you trade the daily time frame, you will have to actually look at the chart once a day. Hmmm. When you look at it that way…
FeedBack Method
Linear Mode
