Dinosaur fights seem impressive but they're just a bunch of matrix multiplications.
We can model dinsosaur fights from the AD&D Monster Manual (1e, of course) as a Markov chain that you run until they hit one of two clusters of final states: one where one has zero hit points, or the other.
The matrices were pretty fiddly to get right but I got there in the end in Mathematica.
I compared the results to two completely independent methods (Monte Carlo simulations in C++ and an experimental Python DSL, all agreed to as many decimal places as you might expect.) so I think I can now reveal the dinosaur league table...