We usually judge moral theories with our intuition. We see what a moral theory says about a hypothetical scenario and see if that agrees with our intuition. People arguing for a particular moral theory frequently give examples of situations in which that moral theory gives an answer that most people's intuitions agree with. People arguing against a particular moral theory frequently give examples of situations in which that moral theory gives an answer the most people's intuitions disagree with.
But a lot of times the examples used are very unlikely situations. That's not a problem for the moral theory. A good moral theory should work in any situation, likely or unlikely. But it is a problem for our intuition.
Intuition is only useful in circumstances it evolved in.
Consider physics. Humans have pretty good intuition with regards to running, jumping, throwing and other things we've been doing for millions of years. But outside our relatively limited experience, our intuition is virtually useless. There's nothing intuitive about general relativity or quantum mechanics.
Which is why I don't think hypothetical problems like the Trolley Problem are useful in determining what makes a good moral theory (though they can be useful in helping us examine our intuitions). Because such hypothetical scenarios rarely reflect our normal experience, so our intuitions don't necessarily apply.
So then, the question is, if we can't trust our moral intuition, then how should we judge a moral theory?