The highly redshifted, sky-averaged (global) 21-cm spectrum from neutral hydrogen encodes information about the universe between the release of the Cosmic Microwave Background (CMB) at z~1100 and the more recent universe we see through our telescopes at z<10. To see this global signal, which is expected to be seen at frequencies of 10 MHz<f<200 MHz, we must look through foreground sources that are 4-6 orders of magnitude larger than the signal; so, foreground modeling is a key aspect of global signal data analysis. While there are theoretical predictions for the shape of this global signal, the most robust possible analysis should not make any assumptions about the shape of the spectrum. We have designed and will present a method, termed the minimum assumption analysis (MAA), for analyzing data from global 21-cm signal experiments that utilize the unpolarized, isotropic nature of the signal to extract any possible spectrum rigorously in the presence of large anisotropic foregrounds. While many common analyses entail the examination of a single observed spectrum, often averaged down from many individually recorded spectra, the MAA takes advantage of the time dependence of the beam-weighted foreground to separate it from the time independent global signal. In addition, as opposed to common polynomial-based models, the MAA uses a beam-weighted foreground model that is specific to the antenna being used. To illustrate the method, we will present results of an MAA run on simulated data, which leads to robust, reasonable results. The MAA can be thought of as a rigorous foreground subtraction technique which allows for any desired signal model to be fit for in the result without the need to fit the foreground simultaneously.