Wright's A grammar of the Arabic language lays out claims for where the stress should be in §§28-31. Most of the claims are connectable (in my experience) to contemporary renditions of Classical Arabic / Modern Standard Arabic (leaving aside the case of Cairene renditions). The one surprising claim is that in words with more than 4 syllables, the stress goes on the first syllable when all following syllables are light (with the well-known complication that final V̄ or VC count as light). Thus [kátabatā, más'alatun, qáṣabatuhumā]. The alternative pattern that I know is that stress is limited on the left to the antepenult.
What evidence if any is there that supports (or opposes) a theory of the location of accent in 4-syllable+ words with all-light syllables in the last three syllables, for Classical Arabic? As far as I know, ancient Arab grammarians did not address the question.