Ok, not to beat a dead horse here, but there's no way that this code is a derivative of the given equation (reattached below). At best, it is also someone's (mis)interpretation. If this code is what's supposed to be interpreted, then that is a different question. Equation interpretation will require a source... is it posted in context on the internet somewhere? Per the forum guidelines, in the 'asking for help' section, post as much information as possible to ensure the best possible answer. Otherwise, this is all an exercise in futility. Really... As you can plainly see, it is 0235 and I am still tring to think this one through. Ever not been able to let something go? LOL! Geesh... Anyway, still unable to test this new thought of mine until Saturday, but what about this? Not flexible, but examines 20 periods. I know ya'll will let me down gently... MB:=mov(typical(),20,s); BarC0:=power((typical()-MB)/MB,2); BarC1:=power((ref(typical(),-1)-MB)/MB,2); BarC2:=power((ref(typical(),-2)-MB)/MB,2); BarC3:=power((ref(typical(),-3)-MB)/MB,2); BarC4:=power((ref(typical(),-4)-MB)/MB,2); BarC5:=power((ref(typical(),-5)-MB)/MB,2); BarC6:=power((ref(typical(),-6)-MB)/MB,2); BarC7:=power((ref(typical(),-7)-MB)/MB,2); BarC8:=power((ref(typical(),-8)-MB)/MB,2); BarC9:=power((ref(typical(),-9)-MB)/MB,2); BarC10:=power((ref(typical(),-10)-MB)/MB,2); BarC11:=power((ref(typical(),-11)-MB)/MB,2); BarC12:=power((ref(typical(),-12)-MB)/MB,2); BarC13:=power((ref(typical(),-13)-MB)/MB,2); BarC14:=power((ref(typical(),-14)-MB)/MB,2); BarC15:=power((ref(typical(),-15)-MB)/MB,2); BarC16:=power((ref(typical(),-16)-MB)/MB,2); BarC17:=power((ref(typical(),-17)-MB)/MB,2); BarC18:=power((ref(typical(),-18)-MB)/MB,2); BarC19:=power((ref(typical(),-19)-MB)/MB,2); Sigma:= BarC0+BarC1+BarC2+BarC3+BarC4+ BarC5+BarC6+BarC7+BarC8+BarC9+ BarC10+BarC11+BarC12+BarC13+BarC14+ BarC15+BarC16+BarC17+BarC18+BarC19; Solution:=sqrt(Sigma/20);

Hey Guys, Bollinger bands are really price plus (minus) the standard deviation of price. This formula looks like standard deviation. Therefore would it not be c +- stdev(c,20)? Also the log should give you a measure of volatility (see Fishbacks ODDS Expert) So how about log(stdev(c,20))?

Cool... I appreciate you being my crash tester (dummy omitted). I don't really understand your revision... :oops: Only because C represents a moving point in a bound range. It looks like your formula keeps a fixed reference point, essentially adding it over and over. I might be wrong. I just don't see it. [quote=Power((Typical()" be something like "Power((((Ref(Typical(),-xx)"? The rest of it is pretty straight forward. I'm assuming that the way I strung all of the pluses together was the wrong syntax. I wish I could have tested it first. Anyway, this is a super learning experience for me. It would be great if there was some sort of a way to declare a variable to be used for looping, but I know that this has already been discussed, so I'm not on a soapbox. (There's not enough room up there with M anyway!

Alpha version 2 or 3 or something: :D D21:=Power((Typical()-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D22:=D21+Power((ref(typical(),-1)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D23:=D22+Power((ref(typical(),-2)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D24:=D23+Power((ref(typical(),-3)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D25:=D24+Power((ref(typical(),-4)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D26:=D25+Power((ref(typical(),-5)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D27:=D26+Power((ref(typical(),-6)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D28:=D27+Power((ref(typical(),-7)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D29:=D28+Power((ref(typical(),-8)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D30:=D29+Power((ref(typical(),-9)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D31:=D30+Power((ref(typical(),-10)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D32:=D31+Power((ref(typical(),-11)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D33:=D32+Power((ref(typical(),-12)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D34:=D33+Power((ref(typical(),-13)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D35:=D34+Power((ref(typical(),-14)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D36:=D35+Power((ref(typical(),-15)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D37:=D36+Power((ref(typical(),-16)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D38:=D37+Power((ref(typical(),-17)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D39:=D38+Power((ref(typical(),-18)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); D40:=D39+Power((ref(typical(),-19)-Mov(Typical(),20,S))/Mov(Typical(),20,S),2); Log(Sqrt(D40/20));

Hi g, I do not know if I am correct or not (I don't have the actual formula in front of me) but the formula sure looks like standard deviation except stdev is norm x bar - x, this formula is x - x bar but it becomes abs when you square it. I do use Bollinger Bands as a visual reference but do not use them for actual trading. There is a lot of history with Bollinger Bands, which are simular to the Process Statistical Control techniques used in manufacturing. There is also an oscillator created from the Bollinger Bands called %B which I like to use. This might be what Imothep was after but %B does not use the log function. Without having some more information it is hard to say what the formula does. Charley Here is the Width computation for BBands since Imothep mentioned it was a formula for width: (BBandTop(C,20,S,2)-BBandBot(C,20,S,2))/(Mov(BBandTop(C,20,S,2),20,S)+(Mov(BBandBot(C,20,S,2),20,S)))/2

There has been some good work done here - but I question some of the logic??? If we had: (x-y)/y we could re-write this as: (x/y)-(y/y) or: (x/y)-1 then: Power((x/y)-1,2) = (x^2 / y^2) - 2(x/y) + 1 What do we get id we sum these amounts continuously over time? How does continuously adding one to the result help? Do we really need it? Adding one just adds distortion. Then: why did we need to take the logarithm? If its just for scaling, that's alright, but we need to understand why each computation is made like it is! Surely the formula: x:=Typical(); y:=Mov(x,20,S); Log(Sqrt(( Power((Ref(x,-0)-y)/y,2)+ Power((Ref(x,-1)-y)/y,2)+ Power((Ref(x,-2)-y)/y,2)+ Power((Ref(x,-3)-y)/y,2)+ Power((Ref(x,-4)-y)/y,2)+ Power((Ref(x,-5)-y)/y,2)+ Power((Ref(x,-6)-y)/y,2)+ Power((Ref(x,-7)-y)/y,2)+ Power((Ref(x,-8)-y)/y,2)+ Power((Ref(x,-9)-y)/y,2)+ Power((Ref(x,-10)-y)/y,2)+ Power((Ref(x,-11)-y)/y,2)+ Power((Ref(x,-12)-y)/y,2)+ Power((Ref(x,-13)-y)/y,2)+ Power((Ref(x,-14)-y)/y,2)+ Power((Ref(x,-15)-y)/y,2)+ Power((Ref(x,-16)-y)/y,2)+ Power((Ref(x,-17)-y)/y,2)+ Power((Ref(x,-18)-y)/y,2)+ Power((Ref(x,-19)-y)/y,2)) /20)); could be re-written thus, for similar results: x:=Typical(); y:=Mov(x,20,S); Sqrt(( Power(Ref(x,-0)/y,2)+ Power(Ref(x,-1)/y,2)+ Power(Ref(x,-2)/y,2)+ Power(Ref(x,-3)/y,2)+ Power(Ref(x,-4)/y,2)+ Power(Ref(x,-5)/y,2)+ Power(Ref(x,-6)/y,2)+ Power(Ref(x,-7)/y,2)+ Power(Ref(x,-8)/y,2)+ Power(Ref(x,-9)/y,2)+ Power(Ref(x,-10)/y,2)+ Power(Ref(x,-11)/y,2)+ Power(Ref(x,-12)/y,2)+ Power(Ref(x,-13)/y,2)+ Power(Ref(x,-14)/y,2)+ Power(Ref(x,-15)/y,2)+ Power(Ref(x,-16)/y,2)+ Power(Ref(x,-17)/y,2)+ Power(Ref(x,-18)/y,2)+ Power(Ref(x,-19)/y,2)) /20); Both of these formula are similar, in that each peaks at the same time, troughs at the same time etc.... just the scaling is a little different. We notice how similar this function is to another formula.... Both of these again are similar to, the simpler: x:=Typical(); Stdev(x,20); Plot each of them and see. Do the more complicated versions reveal more than the simpler version? wabbit :D