aa:=if(c-ref(c,-1)>0,1,0);
bb:=if(ref(c,-1)-ref(c,-2)>0,1,0);
cc:=if(ref(c,-2)-ref(c,-3)>0,1,0);
dd:=if(ref(c,-3)-ref(c,-4)>0,1,0);
ee:=if(ref(c,-4)-ref(c,-5)>0,1,0);
ff:=if(ref(c,-5)-ref(c,-6)>0,1,0);
gg:=if(ref(c,-6)-ref(c,-7)>0,1,0);
hh:=if(ref(c,-7)-ref(c,-8)>0,1,0);
ii:=if(ref(c,-8)-ref(c,-9)>0,1,0);
jj:=if(ref(c,-9)-ref(c,-10)>0,1,0);
kk:=if(ref(c,-10)-ref(c,-11)>0,1,0);
ll:=if(ref(c,-11)-ref(c,-12)>0,1,0);
if((aa+bb+cc+dd+ee+ff+gg+hh+ii+jj+kk+ll)/12<opt1,1,0)

The above (long entry) test is running on my system, with 3 more optimizations for exit, short, buy to cover, and one for a trail stop. Total opts is 32k, and it ran overnight completing only 25% of the test. The attempt is to trade based on # of days up in the last 12, any help appreciated on code correctly identifying the condition, or to speed me uup!

P.S.

c-ref(c,-1)>0

has two operations to evaluate; MS will do the arithmetic, then it will do the comparison to return a TRUE (1) or FALSE (0) answer.

Compare this with:

c>ref(c,-1)

which will have the same result, but completed in less processes.


The difference on a single operation is negligible, but as you have discovered, if you have to do several thousand-million or more passes, then the time taken to do these extra computations becomes very significant.


Remember, MS scripts are not compiled or optimised for execution; it's up to the author to write well.



wabbit [:D]