The correct way to view the problem is as follows. For each source, in each bin, there is some (unknown) expected number of events . The prediction for the number of data events in a bin is not Equation but
From each the corresponding is generated by a distribution which is in fact binomial, but can be taken as Poisson if (which is indeed the case, as our problem is just that a large number of total events gives a small number in each bin.)
The total likelihood which is to be maximised is now the combined probability of the observed and the observed
and we want to maximise
The estimates for the (which we want to know) and the (in which we're not really interested) are found by maximising this likelihood. This is the correct methodology to incorporate the MC statistics: unfortunately it consists of a maximisation problem in unknowns. However, the problem can be made much more amenable.