2 different ways - theoretically and practically.
Theoretically: Using density, infill percentages, known buoyancy values for the T200s and cable, and the volume displaced of water and air (inside the pressure vessels) I could come up with a rough estimate of net buoyancy.
Practically: After having a rough idea of the amount of buoyancy the vessel had without the buoyancy foam, I placed the assembled vessel in the pool and dangled it by a digital hanging scale. The weight on the scale + the desired net buoyancy told me how much buoyancy I needed to add.
Balancing. Also, it is important to balance the vessel front to back and side to side to reduce the idle of the thrusters (assuming you want even listing). This was done through a volumetric/density analysis of the front and rear specific components and taking a moment about the centerline of the vessel and setting it equal to zero.
Having a 1:1 3D model really aids in this process as you can shoot relatively accurate measurements and volume is automatically calculated for you for most of the ROV (you will have to estimate electronic components).
Having a positively buoyant vessel is easier than trying for neutral, and it adds a layer of survivability to the ROV. Either way, you should be able to pull from the tether in the worst case.