I'm going to throw my 2 cents in here, for what they are worth.
In late 2003, I directed a project called PPC (Pokemon Project Creation) that I basically intended to make a new metagame out of the hundreds of fake pokemon I had made throughout the years. Having access to RSBot's source code (a text-based pokemon battle simulator for Ruby/Sapphire over IRC, for those who never used anything like it), I actually coded in a lot of new things. In a span of three months, the project ended up with 137 new pokemon, 95 new moves, 62 new abilities and 3 new types. Most of them by me, but I had some other contributors. I made a website in PHP to document everything that I had done (code that I recycled one year later to make Smogon's first pokedex, ironically enough).
Despite that, nobody ended up playing it but 4 or 5 regulars. Part of the problem is that the community was nowhere as big as it is now; furthermore the lack of nice images (a visual always helps). But the real problem was accessibility. The existence of a full game that you can play to get used to all pokemon excuses the complexity of its simulators. People would look at what we had done and would not bother to learn it - they have to know what the new moves and abilities do, learn the new type chart, they have to pick moves, pick EVs. Competitive Pokemon is not very "pick up and play" without its RPG. Furthermore, I would say that whenever you stray from the official game, people get uneasy - it's just not Pokemon anymore, and it is Pokemon that they want to play. It's a bit like adding your own made up character to Smash Bros. Sure, it might be fun, but it's a "fake" and will always be considered as a curiosity rather than something worthy - nobody will use your character in "true" competitive play for sure. Same here. If you make "fake" pokemon, few people will take your "fake" metagame seriously, even if it is actually better than the "real" metagame. Also, in general, people care about their fake pokemon, not so much the fake pokemon other people made and that is why legitimacy is a very big deal. Your very community-centric system and Smogon's backing will help you a lot with that, hopefully it will be sufficient.
Now, I'm not saying it is not worth it. It looks like a very fun project and I don't think you necessarily want a very big following. But what I would like to say is that there are two viable paths:
A) Keep doing what you do now: a slow increment to the existing metagame. Adding carefully without breaking anything, so people can taste the novelty without getting lost in it. A viable competitive scene will develop with your additions and I think most people will give it a chance, if only out of curiosity, and they'll come back periodically. Smogon's backing will help you a lot by providing constant visibility.
B) Change the core mechanics. The type chart, etc. But if you do that, realize that there is no going back, suck it up and change everything. Redo the type chart completely and by completely I mean from scratch with probably only half of the existing types. Radically simplify the stat system so people can run the math mentally. Flush Nintendo's pokemon because there's no way in hell you're fixing all of them. Wipe out both DVs and EVs. Start out with at most 50 critters. And once you've cleaned house, do yourselves a favor and rename the project to not use the concept of Pokemon anymore.
Basically, if you modify the core mechanics/type chart even in small ways it will completely crash the current metagame. If it does not render it unusable, it will frustrate users because they have to "relearn" the whole game. BUT if you change everything then it pretty much becomes a new game and then people don't have to "relearn", they just have to "learn" and the psychological barrier is a lot lower. If you remove the existing 500 pokemon and most of the math, the game will be very easy to pick up and can still be quite fun. If you're clever, I don't think you lose a lick of competitive value either.
So that's the two ways I think can be successful: slowly incremental or completely radical. Go in-between and you'll have internal problems (because everybody has a strong opinion about, say, the type chart) or problems to motivate people to try it out because it's too different yet not different enough. Personally, if I ever revive PPC, I am going for B). But that's just me and it might be a lot harder to do properly.