You know what? I just got this game (it only came out in Europe quite recently), and I'm a moderately incurable routeplanner.
My current thoughts about TAS tricks (for the main storyline) are as follows:
- The main factors in how quickly you can clear a battle are how much damage you can deal, how many enemies you can include in the AoE of each attack, how many warriors you can deploy (generally speaking you can only viably damage enemies on your own turn; the Metal Mirror and confusion exist, but the Mirror does hardly any damage and I haven't found a confusion source in the entire first half of the game, reliable or otherwise), and how long you can keep those warriors' Pokémon alive.
- The exception to this is banner capture battles. I imagine these are much easier for TASing; in nearly all of them, you can either luck-manipulate, or else manipulate via other means (typically by using attacks that trigger special properties of the battlefield), for the enemies to be away from the banners, then you just take them yourself and defend the banner (not normally too hard if you can stay alive).
- The main factor in how much damage you can deal is type matchup. If you're hitting a 4x weakness (which is often possible), even a fresh recruit is going to do as much or more with their attack as one of your main characters would do with a neutral attack. Meanwhile, the only time I've seen a resisted move do enough damage to even be relevant is with Rock Wrecker, which is really really high damage. The main implication of this is that it's typically going to be worth using fresh recruits and rotating the boss-fighting party constantly rather than trying to maintain a small party and using that, like is the best strategy in the main series games.
- The main implication of that is that you need to be able to recruit quickly. There are two efficient ways to do this, presumably both would be used. The first is to delegate castles to Search; although I find this to give me not nearly enough control to satisfy me on a casual playthrough, it seems perfect for speedrunning. (In general, I'd recommend that Delegate is used everywhere it's legal, = every castle where the main character isn't as soon as the option becomes available.) Mining is probably the fastest way to waste the main character's month, if that ever becomes necessary, although ideally it wouldn't be and you'd just send them to a new boss fight each month, and winning them all somehow. The second is to recruit from the enemy army in boss fights; this can't be done in some battles, but can throughout the first half of the game. This is especially desirable for three reasons: you have to fight them anyway; the "special warriors" often come with Pokémon that higher level than the ones you currently have, particularly powerful, and with good or perfect links; and it has a snowball effect, with the more special warriors you have the easier it is to obtain more (you have to last-hit with Eevee or Jigglypuff within the first 4 turns of a battle until you have other special warriors; experienced Pokémon players may notice a flaw in the type coverage there, and despite their forced inclusion in every battle they won't be doing meaningful damage by halfway through the game).
- The main problem with this recruiting is that new recruits are very hard to keep alive. AI manipulation is one method that can work quite well, especially if you can somehow manoeuvre the opponents to be incapable of moving or attacking and also in range of all of your Pokémon (this is tricky, but hilarious when it works and also very fast). Sweet Song is very overpowered early-game, being a free Super Potion on everyone, but ideally you wouldn't have to spend the time it takes, and it falls off later on. Manipulating misses seems like a better strategy (especially in the Pugilis fight where some of the enemy Pokémon can deal major damage to themselves by missing), but hit chances in this game are often above 100%, so it'd take a concerted focus on Speed and Charisma to make that possible, which might be different to combine with the other goals. Misses also have faster animations than hits (although inability to attack is faster still). And holding enemies in place is worthwhile because movement is quite slow, especially if it changes elevation significantly.
- It's not possible to do all that much to get more mileage out of your attacks. Hitting from behind is useful to get extra damage; hitting multiple enemies (especially if they share a weak point so you can hit them all for supereffective damage) is better but few Pokémon can do it (the main examples I'm thinking of are the hit-two-panels Water and Poison attacks, which sadly often do meaningless damage, and Vine Whip which is awesome); and Abilities and Warrior skills can give noticeable boosts sometimes (Run Up is pretty reliable, for instance, and Celebrate is awesome until it gets let down by being attached to Eevee; note that it always triggers the first time in a turn, but will never trigger twice in a turn).
- Your starters are very good early, but fall off late. Your main character has Top Speed (good for speedrunning because it can sometimes let you get into attack range the turn before you otherwise would); Celebrate (awesome both for speedrunning and for regular play); and Quick Attack (there had to be a downside somewhere, right?). Oichi, who is mandatory for storyline battles, is pretty much the character of Good Abilities With Painfully Slow Animations (apart from Sweet Song, which is not that slow an animation, just you'd prefer not to have to use it at all). This is to the extent that you'd never consider using her in a speedrun unless Sweet Song were absolutely vital (which is possible), except that she's a forced character, boo. (Which raises the interesting question of whether you should actually do anything with her or just leave her sitting still and doing nothing; my guess is that acting and sucking up the animations is faster, but I don't know.)
- In general, anything that increases Range will let you enter combat one turn earlier, but not do much else to help. That one turn is pretty valuable, but can often be made redundant via luck manipulating deployment locations, especially on non-boss battles. (Although you shouldn't be fighting many of those, except in forced tutorial battles; pretty much anything else is faster, and if you need the results from the battle, delegate and let the AI handle it behind the scenes.)
