Multiplexing: Signal to Noise Compare

I wanted to do a S/N comparison between the hadamard (~50% duty cycle) and signal average (~1% Duty Cycle). The counts recorded in the Signal Average datasets are few and far between and therefore, I believe it is appropriate to use Poisson statistics where S/N is measured as square root of the number of counts. For the HT datasets, I measure the noise as (3*stdev of noise region) and signal as the Gaussian fitted amplitude.

Using this approach, the S/N of the signal average datasets always appears higher than that of the HT datasets; this should not be the case from just a pure visual inspection of the data. What is the best way to compare S/N for these two different modes of operation?

Are there any other ways of doing a fair comparison between the HT and signal average mode?
HT_GraphCopy.pxp (99.7 KB) SA_GraphCopy.pxp (37.26 KB)