The standardized numeraire is essentially a minimum-variance basket of currencies similar to the SDR basket or China's reference basket. It is as close as I can get to a mathematically tractable Ricardian IMV. To derive, I take the right-hand Frobenius vector of the positive reciprocal matrix of exchange rates, weighted by monthly trading volumes. The Frobenius root over the trace of the matrix gives an indication of the level of efficiency for the measure.so rather than use the base currency per se, you use an intermediate representation normalized by your "standardized numeraire". I'd be interested to hear how you do that...
The standardized numeraire avoids or at least minimizes the endogeneity problem due to omitted variables (e.g. EURJPY rate may be influenced by movements in USD or other currencies), and also allows more stable and consistent inter-temporal and inter-locale comparisons....and why you don't just use raw currency conversion?