So, it seems likely that the oldest amphidynamic pattern was é/ó-ø-s ~ ø-é-m̥ ~ ø-ø-és (nom ~ acc ~ gen (all amphidynamic nouns are in fact masculine). The newer hysterokinetic and amphikinetic patterns are offshots from this, with various further developments. Therefore, I'm going to take Methinat from a region that never underwent those developments, to make things easier (although, in the long run, I don't think it much matters). Now, this pattern does itself have an interesting two-fold development in Methinat. The three major developments affecting this are: appearance of ə from syllabic consonants; generalisation of the nominative stem to the accusative; generalisation of ə to apophonic zero-grade syllables. These three developments, ordered chronologically, yield the following two patterns: é/ó-ø-s ~ é/ó-ø-əm ~ ə-ø-és (in stems whose suffix does not contain a syllabic consonant in the nominative); and é-ə-s ~ é-ə-əm ~ ə-ə-és (in stems whose suffix does contain a syllabic consonant in the nominative). Thus, we find *déh₃-tr̥-s ~ *dh₃-tér-m̥ ~ *dh₃-tr-és "giver" > dóʔtərs ~ dóʔtərəm ~ dəʔtərés on the one hand, and *déyus ~ *dyéwm̥ ~ *diwés "sky, sky god" > déyus ~ déywəm ~ dəywés.
The proterodynamic pattern was clearly é/ó-ø-Ø ~ ø-é-s (nom-acc ~ gen (all proterodynamics are in fact neuter)). The unique s-stem ablaut is a product of later developments. Unlike the amphidynamic ablaut, proterodynamics all develop the same way, to é/ó-ə-Ø ~ ə-é-es, as in *dór-u ~ *dr-éu-s "tree" > dórəw ~ dəréwes. s-stems, however, all gain a stem alternation between -s- in the nom-acc and -h- in the oblique. Thus: *nébʰ-s-Ø ~ nbʰ-és-s "cloud" > *néβs ~ *nəβéses > néfəs ~ nəféhes. This adds them to the ranks of inherited heteroclites (the numerous r/n-stems, and the occasional l/n-stem). An interesting case is *gʷenh₂- "woman". Szemerényi's Law gives *gʷḗn ~ *gʷnéh₂s, which could easily survive as gḗn ~ gənáʔas, with a synchronically irregular nom-acc sg.
The static nouns are awful to try to reconstruct, but I think it likely that there was a class with ó/é ablaut, though not ḗ/é ablaut. That "Narten" inflection never worked for me. There may have been a class with no ablaut whatsoever, but that doesn't sit right with me either (though, admittedly, it is more likely than "Narten" inflection, given that the two mobile paradigms could have either o- or e-grade roots in the nominative, so it's not out of the question that the static paradigm could too). Anyway, based on what we would expect based on the mobile paradigms, they would probably have this pattern: ó/é-ø-(s) ~ é-ø-(m̥) ~ é-ø-s > ó/é-(ə)-(s) ~ ó/é-(ə)-(əm) ~ é-(ə)-es. A good candidate for a static paradigm is *nógʷʰ-t-s ~ *négʷʰ-t-m̥ ~ *négʷʰ-t-s > nówts ~ nówtəm ~ néwtes.