Who is the intended audience for this? Or just a personal project you want to do 'cuz it's cool?
I'd suggest including a brief history of the DL & RPG gaming - e.g. the era when a ton of RPGs came out on the PSX & PS2, and various well-regarded RPGs came out on SNES with high playership (thanks to emulation if nothing else). And that was kinda ground zero of the DL, and it's a bit trickier nowadays because there's fewer AAA RPGs coming out in general.
I'd also start from an even higher perspective than the likes of scaling: why do this at all, why is it cool. e.g. start with Cloud vs. Squall, who'd win in a fight?! And go for the following methods of voting:
A) Voting based on popularity (Cloud is the coolest and I like him, so he beats everybody!)
B) Voting based on flavor (Suikoden 5 Prince has the Dawn Rune, which is a child of the Sun Rune, so clearly he beats vampires like Hilda. this actually happened as a reminder.)
C) Voting based on in-game effectiveness (Respecting characters like FF6 Relm or Grandia 3 Miranda Dahna; disrespecting characters like, well, a lot of bosses who face extreme broken in-game, or have their tricks usually immuned - SH2 Kato say. or PC status-slingers in games where randoms all suck anyway and bosses immune status.)
D) Voting based on "unique" abilities (e.g. FF6 Edgar gets hype for Tools, but Aeris doesn't get hype for a good magic score to use Materia with)
E) Voting based on "unique" abilities, scaled relative to the cast with math rather than enemies, and scaling all other stats as well (the mathiest and most complicated version, and also the one that the people still remaining at the DL usually use... this can lead into the discussion of scaling.)
For "sample matches" to discuss, I agree that picking 3-5 examples and deep-diving them is a good idea, ideally starting from some that are cut-and-dry in a manner that doesn't require any math to understand, and then ending on ones that are truly close and down to interp splits or random chance. Also focus on games with extremely high playership.
A few suggestions:
* Tidus vs. Ephraim. (Also happened in the DL. At first glance: Ephraim destroys Tidus! Oh wait, Caladbolg has Evade & Counter, never mind. Can hypothetically win on counterattacks, but not after 5x Cheer and possibly Overdrive spam. Can even mention Haste & Slow possibly interacting with double attacks.)
* Some Persona 4 character vs. Somebody who has an elemental attack that would hit weakness. Yukiko vs. FF13 Lightning or something, anything works. Should be a very understandable example of something cool that doesn't require checking the math. Can even mention the P4G version issue which has Evade (element) passive skills and the like for an example of a version split.
* Some final boss vs. final boss slugfest, pick your favorites from the stable of Godlike champs. This one ends up mathier, but can be an excuse to talk about boss scaling.