Bollinger Bands (Part IV of IV)

03/21/01 10:41:09 AM

by Dennis D. Peterson

Bollinger bands are one of the most successful indicators available today.

Security: $compq
Position: N/A

I started this series by saying one of the most useful aspects of Bollinger bands was using the contractions in volatility to signal a trend reversal. Bollinger defines a squeeze, which is the same as a contraction, to occur when the 20-day volatility sigma is one and a half below its six-month average. In other words, when the current standard deviation, defined to be the current 20 days (default period for calculating sigma) is one and one half times below the norm, defined as the six-month average of sigma, a useful contraction has occurred.

Since I use market breadth indicators in my trading, I created a slight variation of that theme. The results I am going to discuss keep me interested in this approach. As a general rule, volume leads price. So I created the standard deviation of Nasdaq up-volume minus down-volume. I then created the moving average of up-volume minus down-volume, added two sigmas of up minus down volume to the moving average, and subtracted two sigmas of up minus down volume. The result is a Bollinger band type of indicator (shown in the top chart of Figure 1). The single red arrow is an anomaly, which I will address shortly. What can be seen is that as the top chart contracts "significantly" the bottom price chart of the Nasdaq composite changes price direction. I will also get to "significantly" in the next few paragraphs.

Figure 1: Nasdaq Composite (bottom chart), true/false condition for a volatility comparison (middle chart), and Bollinger-like bands for Nasdaq up- minus down-volume (top chart). Blue arrows correspond to when volatility comparison is true (a value of one in middle chart).

Graphic provided by: MetaStock.

Graphic provided by: Data vendor: eSignal<.

The Metastock formula for the upper chart is:

voldev:=Stdev((Security("online:$uvolq",C)-Security("online:$dvolq",C)),12);
middle:=Mov((Security("online:$uvolq",C)-Security("online:$dvolq",C)),12,E);
upperband:=middle+2*voldev;
lowerband:=middle-2*voldev;
upperband;middle;lowerband

It calculates the standard deviation of Nasdaq up-volume minus down-volume over the last 12 trading days and then takes the exponential moving average for the last 12 trading days of up- minus down-volume. Finally, I calculate the upper and lower bands and display all three.

I chose an exponential moving average because I wanted to minimize lag and because lag can only help in trend-following systems, such as MACD, to keep whipsaws down (by reducing exits). I chose 12 days because the Nasdaq is a highly volatile market and from my experiences in high-volume stocks I find 12 days to be an optimal number when using Bollinger bands.

Rather than create a trading system, I showed the intermediate step I usually go through to create a true/false indicator to see how well the indicator will do. What I saw from the top chart was that about one trading week, five days, would show a useable or "significant" difference in standard deviation. As a result, I created the following Metastock formula for a true/false condition based on standard deviation and a moving average,

voldev:=Stdev((Security("online:$uvolq",C)-Security("online:$dvolq",C)),5);
ave:=Mov(voldev,12,E);
1.5*voldev
which uses Bollinger's rule that sigma be 1.5 less than the moving average, but with a couple of twists - namely using a 12-day moving exponential average and volatility that is calculated for the last trading week rather than the last 20 trading days.

The result is shown in the middle chart of Figure 1. I looked to see when the condition was true (shown by the indicator going to one in the middle chart of Figure 1) and annotated the Nasdaq price chart with blue arrows when the middle chart showed true. I immediately noticed that the peak in September had been missed (indicated by red arrows in Figure 1). When I looked at the top chart I could see that there had been a squeeze but the squeeze, hadn't been seen by the test. The solution? I reduced the voldev from five days to four and the true condition appeared and not much more.

Summary:
Bollinger bands are used in conjunction with other indicators. RSI often works well. To use Bollinger bands and another indicator, you need to establish an indicator threshold at which upper and lower band walkers will no longer walk the band edge. In other words you are looking for an indicator that will show a divergence with the upper or lower walker. The default for Bollinger bands is 20 days, and this will work in many cases. To see if you need to adjust the period, look for a major or intermediate peak (or valley) and then adjust the band by using a period so that after the first correction the moving average becomes support or resistance. You also want most, but not all, of the values to be contained within the bands.

Stocks and indexes go through periods of contraction and expansion in both price and other values. It is the pauses in uptrends or downtrends that give rise to the contractions in volatility. Using Bollinger bands in conjunction with another indicator can result in equity growth that is worse than just using the indicator by itself, especially if you are taking long positions only. The tradeoff is often a more stable growth, which allows larger capital commitments with less drawdown risk.

Dennis D. Peterson

Market index trading on a daily basis.

Title:	Staff Writer
Company:	Technical Analysis, Inc.
Address:	4757 California Ave SW
	Seattle, WA 98116-4499
Phone # for sales:	206 938 0570
Fax:	206 938 1307
Website:	www.traders.com
E-mail address:	dpeterson@traders.com
Traders' Resource Links
Charting the Stock Market: The Wyckoff Method -- Books
Working-Money.com -- Online Trading Services
Traders.com Advantage -- Online Trading Services
Technical Analysis of Stocks & Commodities -- Publications and Newsletters
Working Money, at Working-Money.com -- Publications and Newsletters
Traders.com Advantage -- Publications and Newsletters
Professional Traders Starter Kit -- Software