Thanks Graham. Neat little solution, and much faster!
--- In [email protected], Graham <kavemanpe...@...> wrote: > > Here is another possible solution that should work > > RangeTest = param("RangeTest", 1, 1, 4, 1); > array = c; > Range1 = Array > 0 AND Array <= 1.1000; > Range2 = Array > 1.1000 AND Array <= 1.2000; > Range3 = Array > 1.2000 AND Array <= 1.3000; > Range4 = Array > 1.3000; > > buy = varget( "Range"+rangetest ); > > > -- > Cheers > Graham Kav > AFL Writing Service > http://www.aflwriting.com > > > 2009/1/18 ozzyapeman <zoopf...@...>: > > Prog - thanks very much for the code. > > > > Very thorough with the traces, and gives me more insight into the > > issues of arrays vs scalars. I will play around with this some more... > > > > > > --- In [email protected], "progster01" <progster@> wrote: > >> > >> Accepting the use of a loop, here is a variation ala the original, > >> that works as intended, but without using a function: > >> > >> { > >> RangeLoopDemo_01.afl > >> > >> by Progster, responding to: > >> http://finance.groups.yahoo.com/group/amibroker/message/134007 > >> > >> } > >> > >> RangeTest = Optimize("RangeTest", 1, 1, 4, 1); > >> > >> Array = Close; // choose an array for division into ranges > >> > >> // ActualRange = 0; > >> > >> // Set some range division values. > >> RDV1 = 3000; > >> RDV2 = 3500; > >> RDV3 = 4000; > >> RDV4 = 4500; > >> > >> // Note: Range1-4 are ARRAYS, not Scalars > >> Range1 = IIf(Array > 0 AND Array <= RDV1, 1, 0); > >> Range2 = IIf(Array > RDV1 AND Array <= RDV2, 1, 0); > >> Range3 = IIf(Array > RDV2 AND Array <= RDV3, 1, 0); > >> Range4 = IIf(Array > RDV4, 1, 0); > >> > >> // Tracing an array is invalid, unless you are in a loop and setting > >> indexes > >> /* > >> _TRACE( "Array: " + NumToStr(Array, 1.1, 0) ) ; > >> _TRACE( "Range1: " + NumToStr(Range1, 1.0, 0) ) ; > >> _TRACE( "Range2: " + NumToStr(Range2, 1.0, 0) ) ; > >> _TRACE( "Range3: " + NumToStr(Range3, 1.0, 0) ) ; > >> _TRACE( "Range4: " + NumToStr(Range4, 1.0, 0) ) ; > >> */ > >> > >> // This works as intended (but somewhat slow) > >> for( idx = 0; idx < BarCount; idx++ ) > >> { > >> // These traces work. Uncomment to see it. > >> /* > >> _TRACE( "Array[" + NumToStr(idx, 1.0) + "]: " + NumToStr(Array[idx], > >> 1.2, 1) ) ; > >> _TRACE( "Range1[" + NumToStr(idx, 1.0) + "]: " + > >> NumToStr(Range1[idx], 1.2, 1) ) ; > >> _TRACE( "Range2[" + NumToStr(idx, 1.0) + "]: " + > >> NumToStr(Range2[idx], 1.2, 1) ) ; > >> _TRACE( "Range3[" + NumToStr(idx, 1.0) + "]: " + > >> NumToStr(Range3[idx], 1.2, 1) ) ; > >> _TRACE( "Range4[" + NumToStr(idx, 1.0) + "]: " + > >> NumToStr(Range4[idx], 1.2, 1) ) ; > >> */ > >> > >> if (Range1[idx] == 1) ActualRange[idx] = 1; > >> if (Range2[idx] == 1) ActualRange[idx] = 2; > >> if (Range3[idx] == 1) ActualRange[idx] = 3; > >> if (Range4[idx] == 1) ActualRange[idx] = 4; > >> > >> // _TRACE( "ActualRange[" + NumToStr(idx, 1.0) + "]: " + > >> NumToStr(ActualRange[idx], 1.0, 1) ) ; > >> > >> } > >> > >> Short = Cover = 0 ; > >> > >> Buy = (ActualRange == RangeTest) ; > >> > >> Sell = BarsSince(Buy) > 3; > >> > >> > >> > >> > >> --- In [email protected], "progster01" <progster@> wrote: > >> > > >> > I think that (at least part of) the issue with the original code > > is that > >> > > >> > Range1 = Array > 0 AND Array <= 1.1000; > >> > Range2 = Array > 1.1000 AND Array <= 1.2000; > >> > Range3 = Array > 1.2000 AND Array <= 1.3000; > >> > Range4 = Array > 1.3000; > >> > > >> > Range1-4 are actually ARRAYS, but are subsequently treated as if they > >> > were Scalars. > >> > > >> > In my testing so far, treating an ARRAY like a scalar seems to > >> > silently use the last value of the ARRAY - leading to no end of > >> > confusion if you don't catch the original error in the first place! > >> > > >> > > >> > --- In [email protected], "ozzyapeman" <zoopfree@> wrote: > >> > > > >> > > Hello, hoping someone can point out the general flaw in logic > >> here. Even > >> > > though I've been working with AFL for six months now, array vs > > scalars > >> > > can still be confusing. All I'm trying to do is pass an array to a > >> > > function that tests it's range. The program then sets a Buy > > according > >> > > to that range. An optimization is performed to find the "best" > > range > >> > > for a given period. > >> > > > >> > > Of course this is not my actual trading system, but merely a test of > >> > > concept for much a more complicated function. > >> > > > >> > > By definition, the Close price has to fall into one of the four > > ranges > >> > > defined in the function. I am using Forex, but any symbol will > > do, and > >> > > would fall into one of the four ranges. Therefore, running the > >> > > Optimization should generate some trades, as the Buy condition > > will be > >> > > true eventually, as we cycle through the "RangeTest" variable > > for each > >> > > bar. But no trades are generated. > >> > > > >> > > Traces indicate that the Close array is not being cycled through. > >> > > > >> > > Shouldn't the following code work, without having to get into a > >> Barcount > >> > > loop? If not, where is the flaw? Any help much appreciated. > >> > > > >> > > > >> > > RangeTest = Optimize("RangeTest", 1, 1, 4, 1); > >> > > > >> > > > >> > > function RangeFind(Array) > >> > > { > >> > > ActualRange = 0; > >> > > > >> > > Range1 = Array > 0 AND Array <= 1.1000; > >> > > Range2 = Array > 1.1000 AND Array <= 1.2000; > >> > > Range3 = Array > 1.2000 AND Array <= 1.3000; > >> > > Range4 = Array > 1.3000; > >> > > > >> > > for( n = 1; n <=4; n++) > >> > > { > >> > > > >> > > RangeN = VarGet( "Range"+ NumToStr(n, 1.0,0) ); > >> > > > >> > > if (RangeN) ActualRange = n; > >> > > > >> > > } > >> > > > >> > > return ActualRange; > >> > > } > >> > > > >> > > > >> > > Buy = RangeFind(Close) == RangeTest; > >> > > > >> > > Sell = BarsSince(Buy) > 6; > >> > > > >> > > >> > > >
