Some feedback:
Early matches are more random, as players don't have the money to strategize. As all non-grudge matches are worth the same in the end, a lot seems to weigh on matches where player agency is low. I'm not sure to fix it, though notably the Evolver and Versatile traits give a massive advantage in the early game.
Gambling odds based on past performance is odd in such a short league. I would suggest applying the matchup-based mathematical odds to them as well. A payout of 133% is equally absurd on either side if the matchup is 35:1, i.e. one side only winning if they roll a 6 and the opponent a 1.
Rare pokémon only pay themselves back if used at least four times. As a player probably only has money for them later in the game, this makes them inefficient unless there are a lot of players or a lot of grudge matches. This makes them, and therefore the Elite trait, dependent on player count. The Fan Favourite trait gives the effect of rare pokémon without the cost. Both of these traits also encourage playing more grudge matches, and also might encourage others not to accept them.
Pokémon only have three traits: type, rarity, and evolution count. As the last of these makes some pokémon far more useful in the late game, some pokémon become disappointing to receive due to the random aspect of purchases. This also balances actual matchups towards common pokémon; only one common pokémon (Onix) doesn't evolve, and only one rare (Dratini) evolves twice, and more than half of the rares don't evolve.
To not make the matchup an arms race of modifying fusions (where the Versatile trait can defeat anyone), there should be a more clear and strict order of actions in a challenge, i.e. clear rules on when you can or cannot modify your pokéfusions. This should favour neither the challenger or the challenged.
There is really no reason to buy promotion before the end, as fans give no advantage until the final ranking.
Overall, I think the game would benefit from a larger player count, even if it brings the problem of exponentially growing match count and player engagement possibly dropping. A possible solution to this could be to have each pair of players play to best-out-of-three rather than just a single round robin. This would reduce the impact of the more random early matches and grow the match count only linearly.