--- In [email protected], "brian_z321" <[EMAIL PROTECTED]> wrote:
>
> 
> Just one, single, solitary, simple question.
> 
> Is BarsSince baby programming or not?
> 
> Brian.
>


Hi everybody,

just my 2 cents opinions.

First, I want to thank Homar, because I've been in trouble with exact 
the same issue before, and now I understand why. Before, I didn't 
came here asking for help, I simply gave up (I understand your 
feelings, Homar, as I too was feeling myself really out of sync with 
AB output).

Second, IMHO – problem is not in how "BuySince" works, or in 
what "baby programming" is.
For me, the problem is neither in understanding what is an array (a 
line or a column in Excel are arrays) but in how BUY works.

Correct me if I'm wrong, but I became aware after some time that BUY  
(as Sell, Cover, Short) is a very special array.

It was especially useful when Marcin pointed me to this table:

http://www.amibroker.com/gifs/bt_regular.gif 

So, trouble (my trouble , at least) is:  BUY can be  == 1 (that is 
TRUE) in "raw signals" and also when you're not buying anything (see 
phase 1) because you have already bought.

This was very confusing to me, even because – once again correct me 
if I'm wrong – in the past  (maybe version 4.50?) the behaviour of 
Amibroker was different: in the past the backtest would open a new 
position every time BUY was true, providing there was equity 
available.

In the past, if one wanted to avoid to buy on the same symbol more 
times with subsequent signals, it was necessary to use

Buy = Exrem (Buy, Sell);

that now – if I well understand – is of no use anymore (even if it's 
still the first example of the "exrem" function in the User Guide).

So, I would certainly agree with the excellent opinions about AB (and 
I too have experience with a certain number of peer software), but 
let's face it: it is not a question of how much dumb I am (or I am 
not) or what is my programming level.
As I like to code in at least two different engines the very same 
idea I want  to test, I see it as clear as a fact: some simple rules 
(like he one Homar pointed out) who are programmed in a second in 
other "backtesting engines" are not so immediate to be coded in AB. 
More time, more trouble.
On the other hand, AB is a real power: it can do things possible only 
in software that cost a multiple of his price, so let's keep on in 
the learning curve…


Reply via email to