Regression Channels can be added to a chart to trade different time frames and styles. A variety of channel constructions is possible once the middle regression channel is drawn, five of which are discussed in ProfitSource’s on-line Contents (navigate to this area from the Help menu). This article focuses on constructing a Linear Regression Channel and next week will address trading approaches using linear regression channels.
Construction and My Channel Lines
A regression line is the line of best fit – it’s the one that minimizes the total distance from each data point to the line. The default data point is typically the closing value for the security. A Linear Regression Channel is constructed by first identifying a start and end date for the regression data. The channel lines are drawn parallel to the regression line using the furthest data value above the regression line (upper channel line) and the furthest data point below the regression line (lower channel line).
I use both current trends and previous trends to generate the regression line, trying to capture as many days as possible. If a downward trend includes a strong acceleration period then a more gradual decline, I generally create two channels or simply use the more recent channel, in this case the more gradual one. There are times when a current channel needs to be re-drawn as new data enters the picture, but in general, I try to avoid this.
Regression Channel Implications
Regression channel characteristics are similar to other trend lines:
- The longer the trend, the stronger the regression channel;
- The more times a regression line has been successfully tested, the stronger it is deemed; and
- An area of previous support or resistance is expected to reverse roles and serve as resistance or support in the future.
In addition to these characteristics, price is generally expected to move from one channel line to another as the trend proceeds. Price may rise towards the upper channel line which serves as resistance. It then retraces to the middle regression line where it either finds support or breaks support. If price is supported the next target is the upper channel line. If the middle channel line fails to support price, it is expected to proceed to the lower channel line which then serves as support.
Strong momentum and volume are expected to accompany price movement through the upper (bullish momentum) or lower (bearish momentum) channel lines. This may result in an acceleration of the existing trend or the reversal of that trend. If price fails to reach its targeted upper or lower channel line, the implications are bearish (upper line) or bullish (lower line). Figure 1 displays a price move out of a long-term regression channel in the spring/summer of 2007 which was considered suspect due to a bearish divergence in momentum.
Figure 1: Three Year Regression Channel with Momentum (10/31/03—10/31/06)
click here for larger view
The channel in Figure 1 could very well have been re-drawn later 2006 with a later start date. Note how price failed to move to the upper channel line throughout 2006. With the benefit of hindsight this seems like the way to go, but I prefer to maintain a longer-term channel with movement characteristics that are consistent, rather than continually re-drawing channels to provide a best fit. In the case of the channel displayed, a second resistance line could be added within the upper channel region. My preference is to monitor momentum and volume, along with previous price movement.
By December 2007, I added a second channel to capture the price acceleration out of the existing channel. I still used the long-term channel for guidance since the move out of the channel lacked bullish momentum and volume, and seemed to be unsustainable. Although examples can readily be supplied showing price that does indeed move from one regression line to another, when you’re drawing these in real-time, it’s not clear whether the channel will prove to be a textbook case. As is the case with other tools and techniques, it takes some time and practice to build confidence with your construction points and decision-making in regards to a new channel construction.
Figure 2: Short-Term Regression Channel with Volume & Potential Resistance (dotted line)
Figure 2 displays a shorter-term regression channel that adds a downward trending resistance line to acknowledge the channel may need to be re-drawn. The less data used to construct a channel (shorter-term), the more you need to assess how appropriate it is.
Figure 3: Multiple Regression Channels
Figure 3 is a view of multiple regression channels including two downward trending channels (red) and a longer-term channel constructed from 2002 – 2004 that was extended to the right (aqua). The support and resistance provided by the extended channel lines still appear to be relevant in today’s market. In my view, it’s a little early to give the same significance to this channel as the 3 year channel displayed in Figure 1, but it definitely appears to be worth monitoring.
The topic will be covered in more detail next week, but for now it’s worth adding here that momentum and volume should be monitored as price moves with the regression channel so you can best gauge whether a support or resistance area will hold or break. Even the perfect set-up can fail – the goal is to put the odds in your favor by identifying those things that support a move in one direction or another.
For more information on Regression Channels, see Gilbert Raff, “Trading the Regression Channel” from Stocks & Commodities, V 9:10 (403-408).
To access other articles written by Clare White, please click here.
Clare White
Contributing Writer and Options Strategist
Optionetics.com ~ Your Options Education Site
Questions for Clare? Visit the Optionetics.com Discussion Board
