QP2'ers: After finding that RSI values in the QP charts data window
and my QP scans weren't quite the same, I've been trying to formulate
RSI for QP, with little luck. I've been working with two basic
formulas for RSI, which turn out to give the same results (though they
look different):
This one is from TASC: RSI = 100(Su/(Su+Sd)) where Su is the sum of
prices on down days and Sd is the sum of prices on up days.
This one is from TA from A to Z, and works out to be the same thing:
100 minus (100 /(1+(U/D))
Where U is an average of upward price change and D is an average of
downward price change.
I arrived at some interesting (but useless) formulas that don't seem
to give the right values. I started from Jeff's formula for Vidya,
which uses upward and downward price change. (CMO is a variant of
RSI).
Here are two scans that I came up with, and anyone's welcome to try
to fix them up (neither one returns values I'm happy with):
output = "rsi.lst";
daystoload=300;
issuetype=common;
exchange nyse,nasdaq,amex;
float updays, dndays, arsi;
integer i;
updays := 0;
dndays := 0;
// develop MAs of up and Down days
for i = 0 to -13 step -1 do
if close(i) > close(i-1) then
updays := updays + (close(i) - close(i-1));
endif;
if close(i) < close(i-1) then
dndays := dndays + abs(close(i) - close(i-1));
endif;
next i;
if (updays+dndays) = 0 then
updays := 0.0001;
endif;
//arsi:= (updays/(updays + dndays))*100;
arsi:= 100-((100/(1+(updays/(dndays+.000001)))));
println Symbol:-6,",", arsi,",", updays,",", dndays;
*************
This next one works on the assumption that RSI is using more periods
to calculate the average change in prices on up and down days:
output = "rsi.lst";
daystoload=300;
issuetype=common;
exchange nyse,nasdaq,amex;
float updays, dndays, arsi;
integer i, d, start, j;
start := 0;
for d = start to 13 step 1 do
updays := 0;
dndays := 0;
// develop MAs of up and Down days
for j = 0 to -13 step -1 do
i := j-d;
if close(i) > close(i-1) then
updays := updays + (close(i)- close(i-1));
endif;
if close(i) < close(i-1) then
dndays := dndays + abs(close(i)- close(i-1));
endif;
next j;
next d;
if (updays+dndays) = 0 then
updays := 0.0001;
endif;
arsi:= (updays/(updays + dndays))*100; //this is equal to formula
below, believe it or not
//arsi:= 100-((100/(1+(updays/(dndays+.000001)))));
println Symbol:-6,",", arsi;
|
