I have a Supercharts formula for "Central Chord" from "Active Investing", by Alan Hull, which contains nested LinearRegValue functions. Recasting this formula into Metastock using the LinearReg function does not provide the exact same result as the Supercharts formula. The results are all lower, by between about 5 to 20%, with this difference not related to the price of the security. Looking for the relationship between these 2 functions, and so how to correct the Metastock formula.

Original Supercharts Formula:

---

round( ((6.854 * LinearRegvalue(close, 13, 0)) + (4.236 * LinearRegValue( LinearRegValue(close, 13, 0), 11, 0) + (2.618 * LinearRegValue ( LinearRegValue ( LinearRegValue (close, 13, 0), 11, 0), 7, 0) ) + ( 1.618 * LinearRegValue( LinearRegValue ( LinearRegvalue( LinearRegValue(close, 13, 0), 11, 0), 7, 0), 5 , 0)) + (1 * LinearRegvalue( LinearRegvalue( LinearRegValue( LinearRegValue ( LinearRegvalue(close, 13, 0), 11, 0), 7 , 0), 5, 0), 3, 0))) / 16.326, 2) My conversion to Metastock:

---

Round(((6.854 * LinearReg(C, 13)) + (4.236 * LinearReg( LinearReg(C, 13), 11)) + (2.618 * LinearReg( LinearReg( LinearReg(C, 13), 11), 7)) + (1.618 * LinearReg( LinearReg( LinearReg( LinearReg(C, 13), 11), 7), 5)) + (1 * LinearReg( LinearReg( LinearReg( LinearReg( LinearReg(C, 13), 11), 7), 5), 3)) ) / 16.326 * 1000) / 1000; Apart from substituting LinearReg() for LinearRegvalue(), I think only change is to accommodate Metastock's Round function (multiply by 1000, Round to nearest integer, divide by 1000, allowing 1 extra dec place rather than figure the round up/round down).

g.stockman I put this formula into indicator builder how to read this? 0.5 oversold, 0.10 overbought? :P

OK, let's try to break it down into bite size portions. I don't know what your charts and values actually are, so you'll have to be the tester... :D [code:1:a68eca342a]a1:=LinearReg(C,13); a2:=LinearReg(a1,11); a3:=LinearReg(a2,7); a4:=LinearReg(a3,5); a5:=LinearReg(a4,3); cc:=PREC((6.854 * a1 + 4.236 * a2 + 2.618 * a3 + 1.618 * a4 + a5)/16.326,2); cc[/code:1:a68eca342a] Your original code is rounding to 2 decimal places??? The thing is, if you try to round the code above you are generating a square wave. Is that what you original output looks like? If so, change the last lines like this: [code:1:a68eca342a]cc:=ROUND(PREC((6.854 * a1 + 4.236 * a2 + 2.618 * a3 + 1.618 * a4 + a5)/16.326,2)); cc[/code:1:a68eca342a]

What time is it there? Shouldn't you be in bed? But thanks for assistance so far. I hadn't been able to Google up any references to Supercharts formulae. See attached screenshots, 3 week & 3 month for OSH. Red: is LinearReg(C,13); Green: is my original conversion Blue: is your conversion, with PREC function Black: is your conversion, without PREC function For 30/09/2005, actual value of curve was 4.06 (as per email - I subscribe to Active Investor, weekly email list as @ close Friday) Curve values: Red: 4.027 Green: 3.748 Blue: 3.79 Black: 3.83544 Good lesson here regarding rounding - it's not on, obviously, although there is only slight variation with your bit-by-bit method. Also, it seems that Metastock has it's own rules of precedence - can be only explanation for variation between my formula and yours - my formula without rounding produced a curve pretty much exactly the same as the one with.

It's reported to be Supercharts, and the Supercharts formula that I've shown here is directly from Alan Hull's Active Investor. But I guess that might have changed, since book last printed 2004. Think I'll shoot email off to Alan Hull. Thanks for all yr help.

Loops? Who needs them... ;) To find the highest LinearReg over 13~52 periods: [code:1:6cd19b51e3]max(LinearReg(C,13), max(LinearReg(C,14), max(LinearReg(C,15), max(LinearReg(C,16), max(LinearReg(C,17), max(LinearReg(C,18), max(LinearReg(C,19), max(LinearReg(C,20), max(LinearReg(C,21), max(LinearReg(C,22), max(LinearReg(C,23), max(LinearReg(C,24), max(LinearReg(C,25), max(LinearReg(C,26), max(LinearReg(C,27), max(LinearReg(C,28), max(LinearReg(C,29), max(LinearReg(C,30), max(LinearReg(C,31), max(LinearReg(C,32), max(LinearReg(C,33), max(LinearReg(C,34), max(LinearReg(C,35), max(LinearReg(C,36), max(LinearReg(C,37), max(LinearReg(C,38), max(LinearReg(C,39), max(LinearReg(C,40), max(LinearReg(C,41), max(LinearReg(C,42), max(LinearReg(C,43), max(LinearReg(C,44), max(LinearReg(C,45), max(LinearReg(C,46), max(LinearReg(C,47), max(LinearReg(C,48), max(LinearReg(C,49), max(LinearReg(C,50), max(LinearReg(C,51), LinearReg(C,52)))))))))))))))))))))))))))))))))))))))) [/code:1:6cd19b51e3] Note that the above 39-nested Max() functions may not run too well in anything less than a Pentium VII 256THz (terahertz) with 512Tb (terabytes) of RAM. For slower machines, try splitting the formula into four. jose '-)