Re: SHARK 32

["DIMITRIS TSOKAKIS" <TSOKAKIS@xxxx>]

Panos Boufardeas asked conditions re:Shark-32, and here is my point.

I was happy in # 2727, when I saw that Shark-32 was a single AFL line
and thought to place in the AFL Library the smallest possible code.
Shark-32 was presented in TAS&C by Mr Walter Downs, (Oct. 1998)
and is a 3-day formation with each day "inside" the previous.
Inside() is a built-in AFL function and the detection of Shark-32 is 
the line

BUY=INSIDE()*REF(INSIDE(),-1)*REF(INSIDE(),-2);

for AA analysis, scan mode.
I change my mind and I will not place the code in AFL Library, 
because the results of this scan are poor and ambiguous.
For ~400 stocks for ~400 days only 50 occurrences with results of all 
kind.
I asked a profit of 10% in the next 20 days of a Shark-32, scanning 
with

a1=INSIDE()*REF(INSIDE(),-1)*REF(INSIDE(),-2);
a2=ref(a1,-20)==1 and roc(c,20)>10;
buy=a1==1;
sell=a2==1;

Only 2 sell in 50 sharks

I asked a loss of 10% the next 20 days, scanning with

a1=INSIDE()*REF(INSIDE(),-1)*REF(INSIDE(),-2);
a2=ref(a1,-20)==1 and roc(c,20)<-10;
buy=a1==1;
sell=a2==1;

The result was 21 sell in 50 sharks, but I think it is mostly 
coincidental. All the under examination period the market was bearish 
and a 10% loss in 20 days was an easy thing.
My personal opinion is that this formation is not useful.
The inside() function as the gapup() and gapdown() are very 
interesting.You may simplify some properties.
Panos, I hope this help.

Dimitris Tsokakis 

--- In amibroker@xxxx, "Dimitris Tsokakis" <TSOKAKIS@xxxx> wrote:
> Well,
> Sometimes Amibroker is great!!
> I was reading Shark-32 formation and I understood that the scan is 
just a line
> 
> BUY=INSIDE()*REF(INSIDE(),-1)*REF(INSIDE(),-2);
> 
> Bravo!!
> (I will put in the library the smallest routine!!)
> 
> Dimitris Tsokakis