Subjective or belief-based probability is the interpretation behind Bayesian statistics. It has been historically opposed (to the point that my university frequentist statistics teacher would mock Bayesians) to frequentist statistics, which is the interpretation behind your words:
All one can say is that the sun rose for X number of days or that we haven’t directly confirmed God to exist and that’s it.
Frequentist probability is in turn the main interpretation behind what is called Classical Statistics. One of its main features is indeed the narrowing of the set of possible outcomes:
classical statistical procedures share the feature that they only rely on probability assignments over sample spaces [..] an important motivation for this is that those probabilities can be interpreted as frequencies, from which the term of frequentist statistics originates [..] classical procedures employ the data to narrow down a set of hypotheses.
Note however that classical statistics might be too restrictive in some scenarios, as hypotheses are discarded when they render the observed sample too improbable, which of course differs from discarding hypotheses that deem the observed sample impossible. This is really well exemplified in the answers to your previous question, Does every possible event have non-zero probability?. The one you marked said no, and it was a frequentist answer, whereas the next, most voted answer said it could happen, it was Bayesian, more flexible if the scenario requires it.
So if you're looking for a philosophy of statistics opposing the subjective interpretation of probability, the most developed as of today is classical statistics, and you should find plenty of references under Problems with the Bayesian approach. However, beware that Bayesian statistics are growing rapidly in popularity, since their flexibility have made them really useful in Machine Learning and Artificial Intelligence.
Finally, a pretty recent (when compared to the other two) proposal is that of Imprecise Probabilities, which are promising since they too differ from Bayesian statistics and are also more flexible than frequentist. Their counterpart in logic would be fuzzy logic. This might fit in your God example interest.