Hi Johnathan

In situations like this I've almost invariably found that there is an explanation for the problem and it turns out not to be a matter of MetaStock being unable to perform a calculation properly. That specific groups of securities return the same result suggests that the cause for the differences you're seeing lies with the specific securities you're using for display purposes. My guess would be that the securities used as hosts have recent extra or missing data bars relative to the broad market data. I've just run a "validity" scan over the most recent 250 bars of 180 US indices and got a result indicating 10 different combinations of missing or extra bars over that period.

Some time back I published an indicator in MSTT that identified data variations between two securities. However, I think it required Volume data and I'm not sure that every index has that. The validation scan should tell you which securities NOT to use as a host if missing or extra data bars is the problem.

Roy

Hi Johnathan

Here's the validity scan. I developed this scan to help overcome problems with the GobalVar DLL when summing historical values across an entire portfolio of stocks. Global variable cells don't automatically align by date and so cell results can only be known to hold data for a specific date across a portfolio when all securities in that portfolio have identical date stamps. Any missing or extra data bars in any security compromises the result. The validity scan is used to select an optimum set of dates and then weed out those securities that don't comply.

{Data Validity Scan}

{This exploration scans a list of securities to find those with optimal/identical data sets. From the various column results select and enable filter values for a second scan which will filter out securities with missing or extra data records. If required the result of the second scan can be saved as a favorite}

{Col A: Day}
ValueWhen(1,Cum(1)=1,DayOfMonth())

{Col B: Month}
ValueWhen(1,Cum(1)=1,Month())

{Col C: Year}
ValueWhen(1,Cum(1)=1,Year())

{Col D: SumMonth}
Cum(DayOfMonth())

{Col E: Sum Days}
Cum(DayOfWeek())

{Col F: Sum Bars}
Cum(1);

{Filter}
{Disable filter for preliminary scan. Use results to set optimum column values for second scan}
{colA=6 AND colB=11 AND colC=2001 AND colD=39308 AND colE=7551}

Filter enabled: Yes
Periodicity: Daily
Records required: As required


When using the Security() function to import data into another chart the Security() function in effect uses ValueWhen(1,date match, data) to "fill" any gaps in the imported data. I've never really looked at what happens when the host chart is the one missing a data bar.

Roy

Hi Johnathan

I'm aware that a problem might only show up in one column - that's why there are 5 separate checks. I look first for common numbers in columns 4 and 5 and then check the start date almost as an after thought. Usually more than one column will signal that something is amiss but it ain't necessarily so. The odds of having a problem when all 5 columns are the same is miniscule, and I've never seen that, but even so I wouldn't guarantee that it's impossible (my maths isn't that good).

Anyway, I hope this tool helps you get a handle on the problem.

Roy