A PR for a change to the infrastructure to put !predict in a reasonable state, or a client-side widget to calculate predict odds, would probably be accepted.
Infrastructure development mana from coredevs is pretty rare and whatever blanket fix has been applied to !predict was probably just the easiest solution dev-side to the issues with the command, rather than a decision that this was the ideal state for the command to be in long-term.
That being said, I don't personally think that !predict is very useful or predictive outside of 1v1 and very small teams.