Not only is it possible to simulate free ligand in solution, but it is necessary. Think of protein and water as two types of solvents. Think of protein-ligand binding as a solvent-substitution event.
Yes, you can simulate only the ligand. It will give you the conformational space of the ligand. However, I am not sure about the stability of the ligand. What do you exactly want to check? If there is any intramolecular hydrogen bonding in the ligand that may fall apart? That kind of things you may check. Otherwise, the ligand shall remain intact; no covalent bond shall break.