Re: ELA CODE -> DUNN'S NOISE TESTER

[Rich Estrem <estrem@xxxxxxxxxxxxx>]

At 02:11 PM 10/7/98 -0700, you wrote:
>
>This comes from the Kaufman interview in S&C bonus issue 98. He talks
about giving markets an efficiency ratio or rank and I guess this is used
in chaos theory as "fractal efficiency". Anyways, there are 3 ways to
calculate this theory....the one with the most detail is to take the "price
change" over a period of time, and then divide that by the individual price
changes over the same period. The ranking goes between 0 and 1. There is
less NOISE in the market and more efficiency as we approach 1, but if the
mrket is wild and crazy, teh denominator swells and we approach 0. I
believe Kaufman's theory is that it is much easy to trade trends in
efficient markets and you can use shorter moving averages...but as markets
get more volatile, you then logically have to use loser, or longer MAs.
This indicator helps quantify the markets.
>
>I'm a butcher of a programmer so I can only figure out how to code this at
a 10 day interval...perhaps Pierre or another top programmer would be so
kind as to make the "extra cool" version that will allow either O,C, H,or L
and any range of days rather than 10.
>
>thanks
>
>Andy
>
>
>
>{DUNN'S NOISETEST}
>{Inspired by Kaufmans "efficiency ratio" and chaos "fractal efficiency"}
>
>value1= absvalue(close - close[1]);
>value2= absvalue(close[1] - close[2]);
>value3= absvalue(close[2] - close[3]);
>value4= absvalue(close[3] - close[4]);
>value5= absvalue(close[4] - close[5]);
>value6= absvalue(close[5] - close[6]);
>value7= absvalue(close[6] - close[7]);
>value8= absvalue(close[7] - close[8]);
>value9= absvalue(close[8] - close[9]);
>value10= absvalue(close[9] - close[10]);
>value11=absvalue(close - close[10]);
>value12=value1+value2+value3+value4+value5+value6+value7+value8+value9+valu
e10;
>value13=value11/value12;
>
>Plot1 (value13, "noise rating");


Andy - this will do any length and any price input:

input:price(close),len(10);
vars:x(0),sum(0);
sum=0;
for x = 1 to len begin
	sum=sum+absvalue(price[x-1] - price[x]);
end;
if sum <> 0 then plot1((absvalue(price-price[len]))/sum,"eff_ratio");


regards,
rich


~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Rich Estrem, Tucson,Az.
estrem@xxxxxxxxxxxxx
~~~~~~~~~~~~~~~~~~~~~~~~~~~~