Quick Update / Backtest Results

The Nasdaq was the relative strength leader early in the session and popped into my "short" zone before rolling over late in the day...

Here's my view of the DOW, the laggard today...

Click Any Chart To Enlarge

First, a little background. I prefer to use the 5 day moving average on a 30 minute chart to capture as much of a trend as possible. As you'll see, there are usually several big moves through any given year that provide the opportunity to ride a trend and score big gains rather quickly. In fact, nearly all of the yearly gains for a trend trader will come from these big moves. Take a look at the current chart of the Nasdaq below. With an entry up around 2290 (highlighted green circle), a trend trader could still be short here with prices now trading at 2126. That's 164 consecutive points in one trade... and counting! While prices moved above the 5 dma in early February, the line never turned up and thus has yet to give the signal to cover...

On the other hand, the DOW and the S&P did give a sell signal over that same period and shook me out of my short positions. That's where this study comes into play. I noted that if I used the 10 dma on the 60 minute chart, I would still be in those short positions. I posted that chart and discussed my thoughts on volatility being a factor in the previous post here.

So I decided to test the 5 dma and the 10 dma in periods when the $VIX was high and match that against a period of lower volatility. For the purposes of this study, I used 24 as my over/under on the $VIX to determine high vs low volatility periods. That number simply worked as the dividing line in the periods I studied. I considered a signal "false" if it was reversed within 3 days and I used the S&P for prices. The goal here is to capture as much of a trend as possible without getting shaken out from a false signal. Here's what I found so far...

The first period is the low $VIX environment from May '06 to Feb '07. Using the 10 dma, I got 9 buy/sell signals with 3 being false. Using the 5 dma, I got 23 buy/sell signals with 8 being false. Not much difference there, as about 1/3 of the signals were shakeouts and obviously, the shorter term moving average gives more trade signals. Here's where it got interesting, the largest trend that the 10 dma captured was 96 consecutive points (1280-1376), while the largest trend the 5 dma captured was 60 points (1320-1380). I'd have to give the low volatility round to the 10 dma and if it wasn't for a "false" sell on 8/14/06, it would have captured 294 consecutive S&P points (1082-1376).

Next, we'll take a look at the high $VIX period from June '02 to April '03. Using the 10 dma, I got 10 buy/sell signals with only 1 being false. Using the 5 dma, we saw 22 buy/sell signals with 2 being false. Note how the number of false signals is comparatively less using either moving averages during a higher volatility environment vs the low volatility environment described above. The ratio went from about 1:3 to around 1:10 The largest trend captured for the 10 dma was on the short side for 174 consecutive points (1074-900). The largest trend the 5 dma captured was also on the short side for 184 consecutive points (1020-836) and if it wasn't for a "false" sell on 6/18/02, the 5 dma would have captured 254 consecutive S&P points (1090-836). I'd have to give the high volatility round to the 5 dma here.

I plan to run these same tests over many more periods to see if I get similar results. I wish I could chart these out for you guys or at least graph the results here and now, but like I said, just getting this data is very time consuming. Maybe I'll save that for another weekend... preferably when there's no Super Bowl!

In summary, the 10 dma captured more of the trend with the $VIX below 24, while the 5 dma captured more of the trend when the $VIX was above 24. While the 5 dma will always give more buy/sell signals, there's very little difference in the ratio of false signals to valid trades when comparing the two moving averages in either a high or low volatility environment. However, there's a significant increase in the number of valid trades vs false signals in a higher volatility environment. Obviously, I'd like to study more data to verify that these results are valid, but it's a start.