Good question. Broadly, philosophers recognize two kinds of truths: correspondence-type and coherence-type. In science, a statement is true if it corresponds with the way things actually are out there. In math / logic, a statement is true if it is logically coherent and follows from a set of axioms / assumptions.

In the case of chess, we can only have coherent truths, because chess is a logical game invented by humans. But some of those truths can be discovered empirically, at least initially, and formally proven only later. That is the case with some mathematical truths as well.