Std Deviation of Complex Calculation

[MadiF]

Hi everyone; I'm trying to produce the standard deviation of the result of a calculation. A very basic mock up of the calc is: 

(((Sum({<[D/U?]={'U'}>}Time) / count({<[D/U?]={'D'}>}[EventID])) / count(distinct ItemID))

/

(((Sum({<[D/U]={'U'}>}Time) / count({<[D/U]={'D'}>}[EventID])) / count(distinct ItemID)) + ((Sum({<[D/U?]={'D'}>}Time) / count({<[D/U?]={'D'}>}[EventID])) / count(distinct ItemID))))

Put more simply, (mean time between events) divided by (mean time between events + mean duration of events).

I tried wrapping this entire calc in stdev(aggr( calc, x)) but I'm getting lots of nulls depending on which dimension are in the chart, which aggregation level/aggr parameter I use, etc. I noticed in the documentation for the stddev function that you shouldn't use the function on aggregations unless those contain the 'TOTAL' qualifier, but I don't want to use TOTAL. I need the result to update depending on which level my end user is at in the pivot table. Is there a better way for me to measure std dev of the result of my calculation? I can't calculate the metric in the script either, as again I need the level to change based on dimensions in a pivot table.