Option Investor
Trader's Corner

Fast Twitch, Smoothed Out

Printer friendly version

For a while, "Fast Twitch" was the exercise du jour at my health club, located adjacent to the Cowboys Training Facility in Irving, Texas. Football players, cheerleaders and the occasional intrepid club member participated. This exercise mode appeared to involve fast movements with no time for recovery, with the apparent goal of building stamina, balance and reaction time. "Fast Twitch" also apparently involved a lot of screaming, certainly on the part of the trainer and, sometimes, the participants.

Having a daughter who endured three knee surgeries after a "fast twitch" gymnastics move tore her ACL, I sometimes wondered if some of those yells and grunts weren't occasioned by injuries. Yet participants in the "Fast Twitch" program at our club would probably vow that no other exercise program could provide both the strength building and cardio punch of that one.

Although electrical engineer and author of several trading programs John Ehlers probably wouldn't characterize an indicator he developed as a "Fast Twitch" indicator, he does claim that it features "essentially zero lag" (Ehlers, "The CG Oscillator"). As any good technical analyst knows, however, an indicator that responds too quickly can give false signals, resulting in injuries to one's trading account. Ehlers claims that his indicator, the Center of Gravity, avoids some of the pitfalls of other so-called "leading" indicators.

In "Hybrid FIR and IIR Filters," Ehlers explains why leading indicators sometimes lead to false signals. One method of eliminating lag, he confirms, is to add a number that represents the slope of prices to the current price in any computations, thus predicting where price will go next. However, Ehler's study of such a method of filtering computations found that this one amplifies the distortions rather than smoothing them.

Instead of employing this method, Ehlers constructed a filter for his indicators that provides smoothing and reduces lag by a substantial amount. He sometimes uses a hybrid filter that consists of a combination of what he terms Finite and Infinite Impulse Response filters. A Finite Impulse Response filter would be used with simple and weighted moving averages, where prior data drops out of the calculation after a certain period. For example, if you're using a 50-sma, the data from 52 periods before is not used in the calculation of the simple moving average. It's dropped out of the calculation. In an Infinite Impulse Response filter, the data isn't dropped. Instead, the prior calculation is used to recalculate the new value. Ehlers says that an exponential moving average is one example of such a calculation.

That's probably about all you want to know about how Ehlers smooths calculations and simultaneously removes the lag or most of the lag. If you want to know more and are mathematically inclined, do a Google search for Ehlers and "Hybrid FIR and IIR Filters." In this article, Ehlers explains how traders might make such adjustments to their own preferred indicators, assuming that their charting program allows such programming changes.

Ehlers employed filters when he constructed an indicator variously called the "Center of Gravity," COG or CG, described in a 2002 article for STOCKS AND COMMODITIES magazine. In this case, the filter was a Finite Impulse Response filter, one type of adaptive filter Ehlers has constructed. Although at least one charting service terms the indicator the COG, Ehlers himself appears to call it CG on his website, so CG it will be in this article. If your charting service doesn't offer this indicator but does offer the ability to code your own indicators, EasyLanguage code for this oscillator is available at www.mesasoftware.com.

As I type, Ehlers may still be tinkering with the CG indicator. He believes that the indicator proves substantially advantageous when compared to others that either lag or are not smoothed sufficiently, but he notes that "the filters have not yet produced the result I seek" (Ehlers, "The CG Oscillator."). He believes a primary advantage this indicator holds over others, however, is that the smoothing he employs clearly pinpoints turning points while the almost zero lag allows traders to move quickly.

Let's look at some charts and see what we find. Ehlers cautions that traders should select a length for the CG that is half the dominant cycle. Selecting too long a length desensitizes the oscillator and selecting too short a length makes it too jittery or "nervous," as Ehlers terms it.

Let's first look at a chart and determine what we believe the dominant cycle and its length to be. Because INTC comprises one of the component stocks of so many indices, let's look at INTC's daily chart.

Annotated Daily Chart of INTC:

The next chart includes the CG indicator with a length of 16.5. Blue arrows have been added at cycle tops with the red arrows remaining at cycle bottoms.

Annotated Daily Chart of INTC:

Let's try the CG indicator with the longer length of 19.5. The first red arrow has been shifted to the left, to indicate the truer beginning of the cycle, as demonstrated by the CG crossover in late July.

Annotated Daily Chart of INTC:

The signals provided here appear to be fairly good ones. It's possible that on the first bullish signal, indicated by the upward-pointing arrow on the RSI portion of the chart in late July, some bullish traders might have entered just before the short-term downturn. Depending on their trading plans, some might have been stopped out just before the continued upward movement that took INTC to a short-term high in early September, but probably only those with tight stops. Being stopped out on a tight stop and then missing a move is just an occupational hazard for traders. It's going to happen occasionally.

However, an obvious difficulty might have appeared to those reading this article. If one knows how long a cycle will be, why would an indicator be needed anyway? Halfway through the allotted length of the cycle, traders could just switch sides.

This is true of an indicator, though, isn't it? When using any indicator, traders must experiment and find a setting that appears to be optimal for the security being watched and the time period being traded. The RSI setting a futures scalper uses might be entirely different than that that employed by an investor looking for a good long-term entry in a favored stock. No indicator will work at the same setting for all traders for all vehicles and across all time periods. Traders still have homework to do once they've settled on a favored indicator.

Another difficulty, perhaps not quite so obvious, might be apparent to those who watch indicators such as the RSI or CCI, both of which are also considered to have less lag than some others, such as the slower-moving MACD. Both RSI and CCI can "redraw" themselves to some degree, appearing to break down or rise through a trendline, for example, only to have the next period's action result in a hooking away from the previous direction. "Just kidding," these indicators seem to say to those who were sure they'd seen that move through an RSI trendline, a move that has since disappeared.

The CG charts above display what happened in the past when all those turning points look clear cut and when few false signals can be found. Did it look that way as events were unfolding?

For example, take a look at early October, when the aqua-colored line was approaching the tangerine-colored one. It's entirely possible that there could have been a crossover that was magically erased after one of those gains in early October. We don't know because we weren't watching at the time. Looking back at an indicator's past results can't guarantee that the same results would be guaranteed in real-time trading.

One particular benefit of CG was evident, if the real-time results replicate the results seen here. Although INTC was trending upward during time period displayed in these charts, the countertrend plays were clearly pinpointed and were reliable. That doesn't often happen, so that benefit alone may justify experimenting with CG a bit. However, until traders satisfy themselves of the reliability of the signals given on their favorite trading vehicle on their favorite trading timeframes, I wouldn't suggest jumping into plays on the strength of this signal alone. I particularly wouldn't suggest taking those countertrend plays. I would, however, suggest using them to guard profits and construct an exit plan once a countertrend signal had been produced.

This article resulted from my own study of various unfamiliar (to me) indicators offered on a new charting service. The information offered here is a result of that perusal and not of my long experience with this indicator, since I have no such long experience. However, if you like what you see here, I suggest you investigate further on your own.

Trader's Corner Archives