TheRumpledOne
6,407 posts
msg #47847
- Ignore TheRumpledOne
modified |
11/6/2006 12:36:37 PM
http://www.insidefutures.com/articles/article.php?id=2257
new approach to trading bands
Well before I had come across Keltner channels I had been experimenting with different kinds of bands looking for volatility based versions that did not use standard deviation. I had long had a problem with standard deviation bands – or in other words Bollinger Bands – since they tend to overreact when a sharp movement is seen. This produced a significant problem at spike reversals since the bands would suddenly widen leaving one side nowhere close to price.
Consider the Chart above where from the left of the chart we can see a period of consolidation where Bollinger Bands have moved sideways and actually narrowed as the consolidation became tighter. Then as price moved sharply lower then level of volatility rose sharply forcing the width of the bands to widen considerably.
While price moved down below the lower band, continuing to remain close to this band, the upper band at Point A actually moved higher. Once the downtrend became consistent then the upper band could normalize and also move lower. However, after the low at 108.96, price was able to rally by nearly 500 points to Point B before price touched the upper band. Apart from the central moving average it really doesn’t give much structure to be able to judge what is happening.
So my challenge was to find or develop a set of bands that provided better short term structure to be able to get a stronger feel of how price is developing.
Standard deviation is a popular method of measuring volatility. However, I needed something else and chose Average True Range.
True Range was developed by J Welles Wilder Junior to represent the real highs and lows of the day to include possible gaps from the prior bars close to the current bars open. This is a tool that was intended more for the futures and equities markets where there is a significant time gap between the close and the following day’s open. In this way True Range is calculated by taking the maximum of:-
1. High – Low
2. The prior bar’s close – Low
3. High – the prior bar’s close
In fact it is very unusual for these gaps to occur in the Forex market since there is no time difference between one days close and the next days open. Thus a gap can only really effectively occur over weekends. Thus for 99% of the time Average True Range in Forex is really the same as Average Range. Still, the information drawn from this does represent volatility in terms of the increase or decrease in average ranges over time. A simple average is then taken of a series of True Range calculations.
Next I needed to construct bands around an average. However, I have never been an analyst that sees much value in simple moving averages. They are slow, lag price movement and I am not convinced that bands should be constructed around a central average since it assumes that peaks and troughs should be equidistant from that average. Clearly they are not.
If we are talking about an average range, then my conclusion was that since we can suggest we know that on average highs are a multiple of average range from the low, and lows are a multiple of average range from the highs then we actually need to use averages of the bars’ extremes. However, I want an average that takes as much lag out of the series as possible.
I therefore decided to use linear regression with the bands being calculated as follows:
ATR Band High = Linear Regression of low price + a multiple of Average True Range
ATR Band Low = Linear Regression of high price - a multiple of Average True Range
ATR Mid Price = Linear Regression of close prices
Using the same chart, this is what they look like:
I use defaults of a 20 period linear regression period and a multiple of 1.764 (0.764 is a Fibonacci number)
It can be seen that when price breaks lower after the consolidation to the top left of the chart the Upper ATR band quickly follows price lower and actually provides good resistance. Equally, in an uptrend the ATR Band Low provides support.
When compared to Keltner Channels it can be seen that the ATR Bands provide a snug channel within which price can move. Keltner Channels in comparison still provide quite loose limits without offering any structure.
ATR Bands are very easy to input into Chart Studio. Open this feature and open a new technique. Cut and paste the following into the area that appears:
indicator Average_True_Range_Bands ;
input period = 20, Width = 1.764 ;
draw line_mid("Mid ATR"), line_upper("Upper ATR"), line_lower("Lower ATR");
vars tmp(series);
begin
line_mid := linreg(close,period) ;
tmp := Width * SMA(truerange(), 55) ;
line_upper := linreg(Low,period) + tmp;
line_lower := linreg(High,period) - tmp;
end.
Following this select “Build” and select “Verify Module” from the top menu bar
You will be prompted to enter a name for this analysis technique. Write in “ATR Bands.”
The Select “Build” again and this time you should see this succeed in the output window at the bottom of the studio.
Then select “Build” again but this time choose “Install Module”
The module will be installed into the User Modules.
You will now be able to access ATR Bands in the charting application via the “set Up Indicators” icon at the top of the chart.
Ian Copsey
My attempt.
|