Hello,
There are two reasons why formula provided below generates slightly different
values at the beginning
a) it outputs one bar later than built-in RSI because of the Ref( C, -1 ) used
that generates response starting from bar 1 not zero.
b) initial (start) value of built-in Wilders() function is not calculated
exponentially but rather as SIMPLE moving average (the
same with EMA).
This is for matching Metastock implementation that people were asking for in
the past.
The averages inside built-in RSI are initialized with zero and use recursive
calculation from the very beginning (bar zero).
If you want to actually DUPLICATE exactly built-in RSI from the very first bar
you need to write loop:
function BuiltInRSIEquivalent( period )
{
P = N = 0;
result = Null;
for( i = 1; i < BarCount; i++ )
{
diff = C[ i ] - C[ i - 1 ];
W = S = 0;
if( diff > 0 ) W = diff;
if( diff < 0 ) S = -diff;
P = ( ( period -1 ) * P + W ) / period;
N = ( ( period -1 ) * N + S ) / period;
if( i >= period )
result[ i ] = 100 * P / ( P + N );
}
return result;
}
Plot( BuiltInRSIEquivalent( 14 ), "RSI 1", colorRed );
Plot( RSI( 14 ), "RSI 2", colorBlue );
Best regards,
Tomasz Janeczko
amibroker.com
----- Original Message -----
From: "gp_sydney" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Friday, July 27, 2007 2:50 PM
Subject: [amibroker] Re: RSI Calculation wrong?
> Graham,
>
> That plot exactly matches what I get when I manually calculate RSI,
> but doesn't exactly match what AB gives. It fairly quickly converges
> to be the same as AB, but is different at the start for some reason.
>
> Regards,
> GP
>
>
> --- In [email protected], Graham <[EMAIL PROTECTED]> wrote:
>>
>> I believe this is same as AB in AFL (plots match), although my
> encyclopaedia
>> uses EMA not Wilders
>>
>> PerRSI = 14;
>> MUp=IIf(C>Ref(C,-1),C-Ref(C,-1),0);
>> MDn=IIf(C<Ref(C,-1),abs(C-Ref(C,-1)),0);
>> Rup = Wilders(Mup,PerRSI);
>> Rdn = Wilders(Mdn,PerRSI);
>> gkRSI = 100*(Rup / (Rdn+Rup) ) ;
>>
>>
>>
>> --
>> Cheers
>> Graham
>> AB-Write >< Professional AFL Writing Service
>> Yes, I write AFL code to your requirements
>> http://www.aflwriting.com
>>
>>
>> On 27/07/07, Chris J <[EMAIL PROTECTED]> wrote:
>> >
>> > Good evening,
>> > I'm using Amibroker to write a system and also the C++ language. The
>> > problem is my RSI value doesn't match up to AmiBrokers. Here is my
> data for
>> > a RSI of 8 periods using GBP/USD;
>> > the values;
>> > 1.9796
>> > 1.9810
>> > 1.9796
>> > 1.9736
>> > 1.9663
>> > 1.9629
>> > 1.9531
>> > 1.9545
>> > 1.9566
>> > differences;
>> > +14
>> > -14
>> > -60
>> > -73
>> > -34
>> > -98
>> > +14
>> > +21
>> > plus_average = .0049 / 8 = 0.0006125
>> > minus_average = .0279 / 8 = .0034875
>> > (plus_average / minus_average) + 1 = 1.1756272401
>> > 100 / 1.1756272401 = 85.0609
>> > 100 - 85.0609 = 14.9391 - my result
>> > Amibroker shows 17.9051 - AmiBrokers
>> > This is the RSIa indicator using the first RSI value and no
> smoothed RS
>> > since it's the first value for both my system and AmiBrokers.
> Driving me
>> > nuts.
>> > Chris
>> >
>> > ------------------------------
>> > Shape Yahoo! in your own image. Join our Network Research Panel
> today!<http://us.rd.yahoo.com/evt=48517/*http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7>
>> >
>> >
>>
>
>
>
>
> Please note that this group is for discussion between users only.
>
> To get support from AmiBroker please send an e-mail directly to
> SUPPORT {at} amibroker.com
>
> For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> http://www.amibroker.com/devlog/
>
> For other support material please check also:
> http://www.amibroker.com/support.html
>
> Yahoo! Groups Links
>
>
>
>
>
