As many members will be aware, various forums have hosted threads on aspects of the adaptive functions that were released with MetaStock version 11. While I've not seen a serious fix for the problematic "cycle" method I have advised past and present MSTT readers of the availability of non-PREV alternatives for almost 40 of the formulas. Non-PREV versions of the remaining formulas are a little harder to create with MFL because of the complex calculations demanded by the underlying functions.

For those formulas for which there is not a non-PREV version available there is still the option of reducing the number of PREVs per formula by two thirds by separating the selection of the adaptive method from the active function output line. In other words the PREV count can be significantly reduced by selecting the adaptive method in one variable and then applying the resulting Periods parameter to just the one function. As the code stands at present each function (with two or four PREVs) is repeated for each of the three possible Periods parameters.

While I've not seen it commented on by others I have noted that regardless of the number of PREVs used by an adaptive function, every function that I've looked at returns incorrect values with almost monotonous regularity. I haven't been able to nail down the cause of this irritating phenomenon but it definitely relates to the "N+PREV-PREV" expression. An alternative expression that appears to overcome the problem, though unfortunately without alleviating normal PREV overheads, is "N+PREV*0". This expression still forces a recalculation of N on every bar without introducing a PREV component to the Periods parameter, but it doesn't appear to generate the recurrent minor errors that are apparent with the original adaptive functions.

Roy

www.metastocktips.co.nz