wow this is impressive. I'm glad to hear from someone who has already worked with the material. I had no idea there was a shrinkage of the material when fired. Oh by the way those ratings looked pretty cool too
Shrinkage occurs through densification. Like: Before sintering the piece is still somewhat porous (The solvent has evaporated, leaving pores) so when you put it in an oven, the pores will fill, which will make it more dense (and a lot stronger) but it'll also have to shrink a bit.
You can get pretty close to fully dense before firing, but that requires a really closely-engineered bimodal particle size distribution (or worse, trimodal).
For perfectly spherical particles, all one size, you can only get to 68% dense, but if you have two sizes, the theoretical max is .68+ (1-.68)*.68 or 90% ish. and the smaller size must be a specific amount (or more) smaller.
----
This is for ceramic powders. For "normal" clays, it's a lot harder to predict the densification and shrinkage, as it's significantly less well defined. (ad harder to characterize)
---
Also, this cold-casting technique looks pretty cool. Howabout instead of metal/ceramic keycaps, we try for epoxy ones first?