The Novice Bear

Using the zero-lag triple exponential moving average (ZL TEMA) cross was the subject of a recent article in The Technical Analysis of Stocks and Commodities magazine.  (Find the code here)  The stated goal in the article was to find a way to get rid of whipsaws in a moving average cross trading system while at the same time not sacrificing response time.  I shared this goal when I was experimenting with using the moving averages of a newly constructed advance/decline line and I thought it might be useful to compare the two methods.

The new advance/decline line (A/D line) was constructed by using the countdown rules of Tom DeMark’s TD Combo.  These are very simple rules that compare the lows, highs and closes of the current bar and two previous bars.  Everytime a buy countdown was recorded a value of -1 was assigned and everytime a sell countdown was recorded a value of +1 was assigned.  The values are then continually summed and then plotted.  What I found interesting was that the line looked very similar to the pattern of the underlying stock but without all of the noise.  Once I saw that the Combo A/D line was acting as a filter, I decided to add a couple of moving averages to it and see if there was any interesting behavior.  What I found was that, like the ZL TEMA chart, there were very few whipsaws.  I then took the chart a step further and created an exponential moving average convergence/divergence indicator (Combo MACD) to more easily visualize the moving average crosses.

Combo A/D and Combo MACD

It is one thing to say that the Combo MACD does a good job of dealing with whipsaws, but how about some real testing?  In order to demonstrate the realiability and soundness of this new indicator, I performed some simple backtests and compared them directly to the exact same tests on the ZL TEMA cross system that was featured in Stocks and Commodities.

The above figure shows a 2500 day backtest of the TEMA cross.  The period was set to 55 and I used the Nasdaq 100 as my sample.  As you can see in the inset, the system does fairly well with a winning % of 37 and gain/loss ratio of 2.2.  Because this comparison is mainly interested in curbing whipsaws, it is important to look at the total number of trades, which is 7830 in this case.

Combo MACD Backtest

Turning our attention to my Combo MACD indicator, I tested the convergence and divergence of the 30 period exponential and 50 period exponential moving averages.  The first thing that jumps out is a dramatic decrease in the total number of trades, down to 1382 in 2500 bars, which suggest that the whipsaws were indeed reduced.  There was also a slight increase in the winning percentage (now 38%) but more significantly, there was a big jump in the gain/loss ratio, now standing at 3.8.  

There was another interesting characteristic of the Combo MACD indicator that can find some utility in gauging the strength of individual stocks and indexes.  Because the indicator normalizes everything in terms of countdown up and countdown down days, the Combo MACD can be compared between any two charts, regardless of it’s price.  As you can see in the column, I have plotted the raw values of Combo MACD.  The indicator gives a very quick and simple sort that can judge the strongest stocks in any given watchlist.  

Conclusions