[R-SIG-Finance] dis aggregate from 3 month rolling average to monthly

[Matt Considine]

You will need at least n actual, consecutive datapoints and knowledge of 
the exact smoothing weights for an n-period smoothed series (or, 
generally, all the actual underlying data for one smoothed point).  From 
there you can deconstruct not only the one corresponding smoothed 
datapoint but also the datapoint that preceded and followed, by backing 
out the missing datapoint from the known smoothed value.  Then you work 
your way up and down the time period to retrieve the series...
Matt