assume that $X$ is Siegel Shimura variety defined over $\mathbb{Z}_p$, you can take its p-adic formal completion $\mathfrak{X}$,and than take it's adic generic fiber $\mathcal{X}$ and get an adic space,you can also take $(X\otimes Q_p)^{ad}$.then $\mathcal{X}$ is a proper open subset of $(X\otimes \mathbb{Q}_p)^{ad}$ (you can see this phonemena already for $\mathbb{Z}_p[x]$) which gives you the locus of good reduction.
it's ok so far but then if you work with say minimal compactification $X^{\star}$ which is proper, then $\mathcal{X}^*=(X^{\star}\otimes\mathbb{Q}_p)^{ad}$,now $\mathfrak{X}\subset\mathfrak{X}^\star$ is open with a complement of small codimension so you can extend some things from $char\, p$ to the whole $(X^{\star}\otimes\mathbb{Q}_p)^{ad}$.
now if you have an abelian variety with a bad reduction over $C_p$ its Néron model is an extension of a tori with an abelian variety of smaller dimensions and hence you get another way to relate this to char p which at cases I see is consistent with the previous way.
So I wonder can we say something about this bad reduction locus geometrically? for example is it possible to somehow stratify it by good reduction locus of Siegel varieties with smaller dimensions?(of course they have smaller dimension than complement but it's ok if we have to use infinitely many of them(indexing by ext of a torus and abelian variety for example)
