GR is built up around an equation (Einstein's equation) that describes the relationship between the properties of spacetime itself, and the properties of the fields and particles that "live" in spacetime. Astronomical observations tell us that on very large scales (larger than clusters of galaxies), matter is distributed evenly across the universe. On those scales, space is both homogeneous ("looks the same at every location") and isotropic ("looks the same in every direction"). So it's natural to look for solutions of Einstein's equations that describe spacetimes that are homogeneous and isotropic. It turns out that all of those solutions have the property that's now called "the big bang": The distance in space between any two curves in spacetime that describe the motion of massive particles in free fall, goes to zero as the time coordinate goes to zero, if we use the coordinate system in which the matter distribution is homogeneous and isotropic. (In that coordinate system, the time coordinate t is only defined for t>0). So all the solutions that approximately describe the large-scale behavior of the universe are telling us that matter was closer together in the past.
Note that the big bang isn't an event in spacetime. (There's no event in spacetime with time coordinate 0). The big bang is the property of spacetime that I described above. The original big bang theory is the claim that the large-scale behavior of matter in our universe is approximately described by one of those solutions. Since then, other theories have been developed, theories in which some entirely different thing is called "the big bang". Unfortunately, I don't know those theories well enough to explain them.