Something to note is that the anime dub seems to pronounce it as /ˈpoʊkeɪˌmɑn/ (poh-KAY-mon/pok-KAY-mahn) whereas in the English* version of Roxie’s gym music the stress is put onto the first syllable.I don't really know how English began stressing Pokemon on the first syllable, but if I had to guess, when we get a made up word we don't know, by analogy we will subconsciously give it a stress pattern. If the word doesn't easily fit into a common pattern then we sort it into the most common stress pattern for three syllable words in English, which is to have the stress on the first syllable.
Funny how much of a difference the localisation team makes.
*for anyone interested, the Japanese version spells out “Dogars” (Koffing’s Japanese name) instead of “Pokémon”