T&L Manip: First Thoughts (Sapphire)

Due to the nature of double fights in Sapphire compared to Emerald, T&L in Sapphire is not only worse, but you could just consider it straight up not solved. This is because the AI will randomly choose one of the opposing Pokémon and make a decision depending on what they targeted. In Emerald, their choice will be made first, and then they'll target based on that decision; this means that if a slave is in kill range you can easily predict what they will do, which cannot be achieved in Sapphire.

The only plausible solution would be RNG manipulating the first turn of the fight. But what do we look for if we do this? Here are my first thoughts:


1. The AI targets Swampert and Swampert is in mid health.
This is very nice as an outcome, as you would get Solar Beam from Solrock and your slave would live for the next turn, which nice for safety: you can no longer get a really bad combination such as Hypnosis + Solar Beam when none of your slaves are alive, although you can get Calm Mind from Lunatone, which is normally manageable if your slaves are alive. Perhaps we can lead with Castform if we anticipate this outcome and Rain Dance turn one?
2. Solrock targets Swampert, whereas Lunatone targets slave.
Same strategy as last time: no bad outcomes from not having a slave that can keep healing Swampert. It would be great to get a good roll from Dive as well in order to 2-shot the Solrock with that move plus Surf.
3. The AI targets Swampert and Swampert is in low health.
This strategy would be kind of interesting: skip X Speeding in order to get a Torrent boosted Surf and simply 2 shot them. Solar Beam + Hypnosis is impossible in turn 2 because you'd be in kill range for Lunatone, so if you ever get targeted by it you will always get Psychic.
4. Double crit 1 turn fight.
This is extremely Utopian, but it's worth mentioning. Perhaps we can find a cluster that has this as one of the outcomes?
5. Double crit 1 turn fight with Torrent.
Same thing as the last outcome: very Utopian. Not as much as the last one, however: if you are in torrent you will no longer have to hit a very hard range on both Solrock and Lunatone. It's hard to imagine a cluster where being at this low HP turn 1 is good in many other outcomes, however.

Now for the bad stuff.

You can somewhat manage what HP you'll be on before the fight by potioning/supering prior to going in. However, this is an important factor: you will have to somehow be at a specific HP for you to get an optimal outcome (of course you can find a cluster of good things to happen for every possible HP amount, but that makes this really hard for both the person who is making the manip and for the runner).
Finding a cluster for good outcomes might actually be harder than it sounds. This comes down to practice, but the fight is REALLY bad. I don't think any less than a 4-frame window should even be considered at all.
Castform sometimes dies to Solrock's attack, which may make some of the outcomes bad. It is rare for it to die to Solrock, though.
This can just straight up not be faster and just safer depending on the outcome you get. Saving + resetting is roughly 15 seconds of timeloss.
The manip might be very hard since you have to time two inputs (even 3 if you want to manip a crit): one for the last textbox before the fight, and the second one for the first textbox of the fight, which determines what first move the AI will use. The third one would be timing when you use the move.

I will post another pastebin with what outcomes I get from this manip sooner or later. Input is always appreciated.