Tips on Technicals - Bollinger Bands®
Moving averages can be expanded to create trading envelopes called "bands." Typically, the average is shifted up by a percentage of the moving average value to create the upper band and shifted down by the same percentage to create the lower band. These "percent bands" have been in use for many years. The theory is that prices in any market are somewhat inelastic, meaning that they cannot stray too far, too fast from current levels before supply and demand effects force prices to halt the move, at least temporarily. Bands try to quantify just how much "too far" really is.
Bollinger Bands®, also known as standard deviation envelopes, were developed by John Bollinger in 1982. Bollinger theorized that the band width of an envelope should be determined by the market rather than by the assumptions of the analyst, as done with percent envelopes. His theory states that a trading envelope's distance from the mean is a function of the market's volatility. This makes sense because a volatile market does have wider swings from its average, even though a distinct trend may be in effect. The bands should expand to account for this, rather than be subject to giving false reversal signals. Conversely, when a market is flat, the bands should tighten so that a breakout can be signaled early.
To construct Bollinger Bands®, a central moving average must be selected. For most markets, especially equities, Bollinger recommends a 20 day average for medium term analysis. The length of the moving average can be adjusted but only when the particular market presents a reason to do so. It is important to note that the average used for the bands is not the same one used for moving average crossover signals. The best way to select an average is to find one that provides support or resistance for the first correction after a market high or low, respectively.
Figure 1 shows a 330 day chart of Merck & Co., a NYSE stock. Note how the 20 day moving average provided resistance for the decline in early 1993. When the trend reversed in April 1993, the first correction of the new rally was supported by the average. Later, when the market reversed lower in June, the average provided resistance to the first correction higher.
Figure 1
Figure 2 is the same chart with Bollinger Bands® added. Bollinger recommends that each band be shifted two standard deviations from the average so that prices remain within the bands about 85% of the time. When prices move outside the bands, the following analysis can be applied.
- Sharp moves after a relatively calm market tend to occur after the bands tighten to the average (volatility lessens). In November 1993 the bands became quite narrow and that was followed by a quick three point move.
- Reason—market participants have slowed their activities and are waiting for the market to tell them where it is heading. Once a move starts, everybody jumps in to trade it.
- A move outside the bands calls for a continuation of the trend, not an end to it. Often, the first push of a major move will carry prices outside of the bands. This is an indication of strength in an up market and weakness in a down market. This can be seen nearly everywhere on the chart.
- Reason—volatility has not expanded yet to compensate for the new trend. Other indicators, such as RSI, can confirm this.
- Bottoms (tops) made outside the bands followed by bottoms (tops) made inside the bands call for reversals in trends. Note how Merck made a new high in December 1992 outside the bands while the next price peak occurred inside the bands.
- Reason—market participants have not accepted the increase in volatility and therefore look for wide price swings to take profits.
Figure 2
As mentioned, a 20-period average with two standard deviations is commonly used. However, Bollinger points out that the interval one is analyzing may require longer or shorter average and deviation calculations. For example, analysts trying to gain perspectives on longer-term periods may find 50 day averages with 2.5 standard deviation bands optimal. Conversely, those studying short-term intervals may find a 10 day averages with 1.5 or 1.0 standard deviations more interpretive.
