Trading Seminars

Serving the Commodity Futures Trading Traders World

Envelopes & Channels

Trading with an envelope formed by bands around a moving average (or around some similar indicator) is a well-known and effective method of smoothing out short-term whipsaws common to most trend-following systems. There are almost as many methods to use envelopes, as there are technical traders. Most present-day software packages provide the ability to display various forms of envelopes surrounding a moving average.

There have been many internet based trading system software tests conducted which show price channels are one of the most effective trading tools available. Perhaps the most well known of these was a channel breakout study done by Frank Hochheimer of Merrill Lynch® more than 10-years ago. In addition to Hochheimer, there are many other well respected commodity traders who have seen the merits of channel trend trading. For example, although best known for his work with moving averages, Richard Donchian is also known for his channel trading using his 4-week rule.

We think channels and envelopes can be used in a variety of interesting ways to serve as the foundation of a profitable trading system. Many such technical indicators can be learned at a commodity futures university and at other trader educational centers, including trade seminars.

We will divide our discussion into 2-sections. The first will deal with price bands that form an envelope around a moving average. The second section will deal with channel-breakout systems.

Section One - Trading With Envelopes

Envelopes can be as simple or as sophisticated as you want to make them. The simplest is a single moving average with a band on either side calculated as a percentage of the moving average. The area within the two bands theoretically acts as a buffer zone that will tend to contain prices, especially if the underlying market is in a trading range. The beginnings of a trend will normally be indicated by a break outside of the bands. Trend corrections or the end of the trend will see prices move back inside the bands toward the moving average.

Another simple type of envelope is one that uses an absolute point value on either side of a moving average. For example, a 10-day moving average in U.S. T-Bonds might be surrounded by bands that represent 1-16/32nds points or $1,500. Normally this is the risk the trader or money manager wishes to take in a trade in any given market, rather than a specific point at which to enter a trade.

The variations of the two types of envelopes described above are almost infinite. For example, the moving average can be exponentially or otherwise smoothed. Or, the percentage of prices that will be contained by the bands can be varied for the long side versus the short side, thus biasing the study in favor of increased volatility in the direction of a trend.

Another possibility is using a moving average of recent highs and lows to define an envelope. The bands are intended to contain and define the random price action of a trading range. Any breakout beyond either of the bands should signal the beginning of a trend, as prices are no longer wandering within the envelope.

A fairly recent and worthy addition to the ranks of channel studies are Alpha-Beta bands and Bollinger Bands (named after John Bollinger, technical analyst for the Financial News Network - FNN). Alpha-Beta bands and Bollinger Bands are statistically defined bands around a short-term moving average.

The computer software first calculates a simple moving average and then calculates a moving standard deviation in parallel with it using the same data series. Bollinger uses band two standard deviations on either side of the moving average.

He explains how two standard deviations will theoretically contain the vast majority of subsequent data. He also points out how the bands rapidly expand and contract with market volatility, making them very sensitive to recent market action.

Instead of being 2 standard deviations away from the moving average, the Alpha-Beta bands can be set at any increment of standard deviations from the moving average. The usual setting for the Alpha-Beta bands is only one standard deviation on either side of the moving average.

The basic concept behind the statistically derived envelope widths is that the volatility of the market being studied is what determines the placement of bands. Using these self-adjusting bands means that volatile markets will automatically have wide envelopes and less volatile markets will have narrower envelopes.

General Envelope Trading Rules

There are almost as many possible-trading rules for envelopes as there are rules for constructing them. The rules are, or should be, based on the idea the envelope contains a significant amount of the price movement of a market and that a move to or beyond one of the bands is aberrant price behavior and should be acted upon.

Traditional trading rules for envelopes are:

1. Enter market when a band is penetrated. This means that a trend may be beginning.
2. Exit and reverse the position when the opposite band is penetrated. (Use Closes)

1. Enter market when a band is penetrated. This means that a trend may be beginning.
2. Exit the market when the same band is penetrated in the opposite direction, or when the moving average between the bands is reached.

Both of these sets of rules ensure that a major trend will not be missed. The first set of rules is basic and results in a pure reversal system. We have an inherent skepticism about reversal systems and prefer the second set of trading rules. In the second set, the bands are also used for entry but the moving average is used for exit.

If prices are within the bands after a trade is closed out, the market is in a neutral zone and there will be no new trades until there is a new breakout. Another reason we prefer the second set of rules is because the theoretical risk on any one trade is reduced to the distance between the band and the moving average instead of the total distance from band to band.

"Optimum" Percentage for the Bands

The "correct" values for the moving average and the surrounding envelope are elusive. The most extensive testing we've seen covers the period from 1960 to 1978. It appears in an article by Irwin and Uhrig in the December 1983 issue of Review of Research in Futures Markets.

The authors used a breakout of the bands for entries and exited when the moving average in the middle of the envelope was crossed. (They used the second set of rules mentioned above.) They then optimized for the best combinations (our usual caveats about optimizing apply). Here are the numbers they found to be most profitable:

Trading Within the Envelope

A technique we rarely see discussed involves using bands as overbought/oversold indicators so that the trading takes place within the bands instead of outside. We and other traders have used this to great success when a market is in a trading range. The trading rules are relatively obvious and simple. Buy when the price hits the lower band. If the trade goes against you, exit on a close below the lower band. Take profit and reverse the position at the upper band, applying the same rules in reverse.

How do you know when the market is in a trading range and when it's trending? An objective method is to use an 18-day ADX. If the ADX is rising, and/or if it's over 25, the market is trending and you're better off using the trend-following envelope method. If the ADX is falling and is below 25, trading within the envelope can be very rewarding.

Section 2 - Trading Channel Breakouts

In addition to envelopes that are defined by distance from a moving average, there are also channels that are defined by high and low points over time. The simplest of these channel methods is a pure reversal system, which is always in the market.

An upper band is formed by the high of the past 10-days, for example. A lower band is formed by the low of the same number of days, with the two bands forming a channel.

The channel will change in width as old highs or lows are dropped and as new highs or lows are made. The system is long when the upper band is penetrated, and stays long until the lower band is penetrated, at which time the long is closed out and a short position is assumed.

Donchian originally popularized this trading system in the 1960's as the Weekly Rule. He used a four-week time frame, buying when prices exceeded the four-week high, and selling when prices dropped below the four-week low

Bruce Babcock published the results of his own test of the four-week channel in his Dow Jones-Irwin Guide to Trading Systems. Babcock found Donchian's method to be reasonably profitable over time, although the drawdowns were pretty breathtaking.

As you might imagine, the risk at any given time can be sizable depending on the size of the four-week range. In addition to the risk on each position, the total portfolio risk would also be very large because the system does not employ risk control stops.

It's worth pointing out, Bruce Babcock's testing results included over $43,000 in losses on the S&P 500. This isn't unusual; we and many other traders have found that the S&P behaves differently than other futures markets.

The Tempus Formula, a very popular and expensive system marketed in the 80's, was basically the same as Donchian's method except that the time frames were optimized for each commodity. After many years of profitable trading, the drawdowns in 1988 were so severe that many users of the formula were forced to stop trading it. In fairness to system, 1988 was a disaster for many trend following methods.

Selecting the Time Values

What is the optimum number of days to use in constructing a channel breakout system? Hochheimer's Merrill Lynch study we referred to at the beginning of this article produced the following optimized number of days for a channel breakout system.

As you might expect with an optimized study, these channels proved extremely profitable over the 6-year period of the test (1970-1976). However, even with the benefit of optimization, only 42% of the trades proved to be profitable. It should also be noted that the drawdowns were very substantial and the 4-day channel in silver produced 1,866 trades (more than one trade per business day).

It is easily possible to create a channel breakout system with values optimized for each market, but in our experience these systems break down fairly quickly. As (The Late) Bruce Babcock has shown in his test of Donchian's method, a single number can work and be profitable for a diversified portfolio. In fact, as we pointed out, if the S&P were not part of the portfolio, the system's profits were excellent.

William Gallacher, in his wonderfully written book Winner Takes All: A Privateer's Guide to Commodity Trading, back tested a 10-day breakout system on ten different commodities over a period of 130-weeks.

Trading results showed this simple 10-day channel produced profits at a respectable rate of 24% annually. (By the way, we don't know if Gallacher's book is still in print, but if you ever see it? buy it. It's one of our all-time favorites!)

Commodity Trading, back-tested a 10-day breakout system on ten different commodities over a period of 130-weeks. The results showed this simple 10-day channel produced profits at a respectable rate of 24% annually. (By the way, we don't know if Gallacher's book is still in print, but if you ever see it -- buy it. It's one of our all-time favorites!)

Our research and experience with channel breakouts, which is fairly extensive, shows 18-days to be a good number, which works well in many commodities over long periods of time.

We're of the opinion almost anything in the range of 10 to 30-days will be profitable over time. The drawdowns will be of different sizes and occur at different times as numbers change.

Reducing the Risk with a Neutral Zone

A creative way to lessen the drawdowns inherent in channel-breakout trading systems without giving up any profit potential was developed by a money-manager of an acquaintance in Southern California. His system uses different time periods for entries and exits. The exit bands for his method are one-half the time period of the entry bands. For example, if the signal to buy soybeans is a penetration of the high of the last 20-days, the inner channel and exit point would be the low of the last 10-days.

This addition to the Donchian system has the advantage of drastically reducing overall portfolio risk. It also takes away the pure reversal nature of the system by creating a neutral zone in which there is no trading. This should have the effect of reducing whipsaws in choppy markets as well as preserving more of the profits because of quicker exits.

reprinted with permission of Technical Traders Bulletin & webtrading.com