I want to make a mobile phone game that has online multiplayer battle support. But I know I don't have the skills to make one with real time interaction on the player's part. So I want to make one where the player has all the characteristics/strategies set up for their character before the battle, as does their opponent. So each player's character auto battles each other rather than be controlled by you and your opponent. You and your opponent still get the feeling of fighting each other (human vs. human rather than computer) since it's your predefined strategy vs. theirs.

Does this sound doable? What problems do you see arising either with the implementation or end result? I know I'm speaking very broad as I've just labeled this a "battler" game, but I wanted to keep it on broad terms for now. One thing I'm worried about is that the game may seem too rock-paper-scissor-like. Strategy x will always beat strategy y but always lose to strategy z. There's no direct player influence since there's no active control decisions during the battle. Another problem that could arise is one strategy becomes the dominant strategy and everybody will want to copy it. How can these issues be oversome?

Rock-Paper-Scissors is unavoidable imo, just you may make it more varietal (something like Sheldon Cooper's rock paper scissors lizard spock) . If you don't mind it taking ages, you may also consider something turn based on a primitive map.

It's a catch-22: You want players to be able to rely on matching specific strategies to specific threats/situations, but you don't want them to be certain of the outcome. Randomization might help here (50% chance that rock beats scissors) at the risk of frustrating players who fall victim to the RNG. Alternately as Unduli suggests you can deepen the relationships, possibly to the point where without careful mapping and strategy guides a player can't divine a dominant strategy without a lot of play experience (at which point the retire the game). My big fear taking that approach too far would be that you end up needing to do what Wizards of the Coast did in using statistics and lots of repeat play testing to balance unintended, cascading consequences of so much complexity; or constant rebalancing updates, as we sometimes see with CCGs.

I don't see a problem with the core idea itself. I can see it working especially if you break the automatic combat into rounds, which can let players try gambits and advance or recover progressively to the end of the fight.

Consider checking out the Smugglers games for a fun, abstract strategy framework that might be similar to what you're proposing.