Back about 2002, I received emailed "LOL" comments when I accidentally typed "recession" channel when I meant "regression" channel. I was accused of letting my bearish bias show.
However, snapped-on linear regression channels, calculated by one's charting service, help to remove bias. They're a simple tool for discovering whether there's a trend and how long it has been in place. They help a chartist discern where support or resistance might lie as long as that trend stays in place. Furthermore, they help the trader determine when that trend might have ended.
While economists might argue whether the Great Recession has ended and I might personally fear that a pullback is overdue, the automatically constructed linear regression channel shown below illustrates that, as of August 8, when the chart was snapped, the SPX was still well inside a rising linear regression channel that began forming in late February or early March. Width requirements for publication required a resizing of the chart. That rendered the chart less legible than is optimal. For example, the candle shadow left behind on 8/7 is not visible. However, the channel's parameters remain clear and that's sufficient for the purposes of this article.
SPX Chart as of 8/8/09:
After the first strong move, prices have tended to congregate along the midline of the channel. A head-and-shoulders formed along the midline from April through June, followed by a downdraft that set up a June-through-July inverse head-and-shoulders, also with a neckline along the midline. That's an important midline. As of August 7, that midline was at about 983-984, but it's now progressed higher, to about 1023.
What can a trader surmise from that chart? The SPX was in an uptrend, but traders knew that. Sometimes snapping a default regression channel on a chart makes clear a trend that the trader hadn't realized had begun, but it would have been hard to miss the uptrend that began in March. There's more to be gleaned, though. For example, we can tell now that a sustained drop below the midline of that channel would move the SPX back into the bearish or weaker half of this rising channel. Such action might promote the idea that the SPX was weakening and vulnerable to a drop toward bottom support again. As long as the SPX maintains values above that rising midline, however, it maintains the potential for a climb to the top of the channel again.
That channel is rather broad, rendered so broad by the first volatile swings that formed it. Later moves have clustered around that midline. Dialing down to a three-month channel might offer more telling short-term guidelines. Again, width requirements for publication render this second chart less than optimally legible, but the new, slightly tighter channel is clear.
Three-Month Chart of the SPX as of 8/8/09:
In this version, the channel's top line, where resistance is presumed to lie, was at 1032.30 at the time the chart was snapped, but was still rising. As long as daily closes were sustained above the channel's midline, a test of the top resistance line remained plausible, and that resistance was indeed to be tested.
Sustained daily closes below the midline, should they occur, would render bulls vulnerable to a test of the channel's lower line. The midline was at about 967.70 at the time this chart was snapped, but a new three-month chart, checked Thursday night, shows the tighter channel's midline now at about 1010. The bullish trader who'd bought the breakout above the midline, anticipating a particular upside target, would be alerted to rethink upside targets and tighten stops if the midline was violated on consistent daily closes. A bullish trader who had been in the trade since March might be willing to ride prices all the way back to the bottom of the channel for a retest while a bull who had bought only after the three-month channel's midline had been closed would probably be considering taking off the position when the midline was violated on consistent daily closes. Thus, these regression channel can help those in long-term and short-term trades make decisions about appropriate stops and targets.
Is the upper channel line always tagged once daily closes are consistently above the midline? Is the lower channel line always tagged once daily closes are consistently below the midline? No, of course not, as the chart shows. For example, the June high failed to reach the top of the channel before a pullback took prices below the midline and all the way to the bottom channel line. Regression channels are like any other type of trendline, channel or band one might put on a graph. They're simple, useful tools that help traders make what-if plans.
Using a favorite lower indicator such as MACD, RSI or CCI can help traders verify overbought conditions as prices test midline or upper boundaries of a regression channel and oversold conditions as prices test support at the midline or lower boundary. Such indicator evidence is also not an ironclad predictor of what will happen next, but that evidence can be useful corroboration of what you're seeing with regard to a linear regression channel. For example, when June's high was bounding above May's, bearish price/RSI divergence developed at the June high. That might have led the bullish trader to cinch up stops, reevaluating the likelihood that prices would reach all the way up to the top of the channel. On its own, without the follow through in price action that was to occur, a bearish price/indicator divergence might not have been enough evidence to cause bullish traders to sell all their longs at that time. However, it certainly would have been enough evidence to tighten stops, just in case, and rethink the likelihood that the SPX was going to climb all the way to the top of the channel on that particular swing above the midline.
If linear regression channels are so useful, what are they and how are they constructed? The "regression" in that term refers to regression analysis, a mathematical term. Prices are used as the input for our charts, but this type of analysis can be applied to many types of sets of "noisy" data. For our price charts over the period being studied, the least-squares method of analysis that you might dimly remember from high school or college mathematics is used to design a channel that filters the noisy price points and constructs a smooth channel. We don't have to remember exactly how to manage those least-squares calculations, since computers do them so handily. IQ Charts suggests that default regression channels work best when the period of time chosen for the channel is similar to the typical trending period for the underlying being charted.
You can draw your own price channels, but snapping on a charting-system-calculated regression channel is less biased. It can turn up surprising information. For example, the periods we've used so far for the SPX are six-month and three-month periods, and they've shown rallies inside rising regression channels. What if we look at a two-year chart?
Two-Year Weekly Chart of the SPX:
Wow. That gives a completely different picture, a longer-term picture into which we must put the context of the shorter-term look provided by the six- and three-month charts. The strong rally we've seen on those charts has served merely to bring the SPX up to top of this longer-term, two-year regression channel. On this chart, prices obviously press up to the top of the descending price channel, as they still do on the weekly chart I viewed last night. It's possible that the channel is just expanding with the price movement, as the longer-term regression channels such as the three-year one have higher boundaries. However, the 14-week RSI was the highest it's been since October, 2007.
What does this prove? Nothing yet, except that the SPX is at a key level, primed for a break out of the top of the long descending price channel or a drop back inside that channel again, perhaps as low as the midline. Having felt the violence of the climb off the March low, many in bullish positions might be startled by this view or by a four-year chart, which shows the SPX pressed against the midline of a longer-term descending regression channel. Having felt the violence of that climb, bulls might feel that the most likely next step is a break out of the upside of that channel, while bears might point to the RSI's corroboration that the prices are at a particularly vulnerable point.
I'd hoped that by the time this article was published a couple of weeks after I'd first roughed it out, we'd have some answers. Unfortunately, we don't yet. These charts predict that it's time for caution if in bullish trades, but no channel or lower indicator can predict the outcome of the test. They just let us know how prices are organized, whether a trend is in place, and when we need to reevaluate our stops and what-if plans. Do some evaluating.