Saying "Treecko depends on RNG to beat Roxanne" is like saying Vileplume is bad because it relies on the 75% accurate sleep powder.Rock throw and rock tomb have 90% and 80% accuracy, respectively, so the chances of 3HKO are really 72.9% and 51.2% when considering accuracy. Treecko's bullet seed averages 7 damage per hit, so it almost always needs 6 hits (rarely 5 or 7) to defeat Nosepass. If Treecko uses bullet seed twice, it can hit 5 times only if both uses hit 2 times, the first hits 3 and second hits 2, or first hits 2 and second hits 3. 3 * (3 / 8) ^ 2 = 27 / 64, or roughly 42.2%. The only way Treecko loses the 1v1 is if Nosepass uses only its STAB and doesn't miss, and Treecko's bullet seed hits less than 6 times when Treecko uses it twice (or if Treecko is level 13, but even that still has a higher than 50% chance of winning). It turns out to be roughly 38.8% if Nosepass spams rock throw or 21.6% if rock tomb. It's obviously not 0%, but I think it's reliable to be considered a good matchup. Brawly is not a good matchup, though. Also why are you using absorb when bullet seed has more PP and is guaranteed to be stronger (because damage works weird for multi-hit moves used by low-level Pokemon)? Sceptile doesn't beat Drake nearly as easily as stuff like Swampert, Gardevoir, Hariyama, Gyarados, Sharpedo, or Starmie, but high speed and dragon claw are still better than what most other Pokemon can offer. A level 49 Sceptile can outspeed Drake's entire team (ignoring dragon dance), defeat Shelgon and either Kingdra or Flygon, and possibly help revenge-kill Altaria or Salamence.
This is by no means saying that Sceptile should move up from C. As one of the people who
tested Treecko last year, I totally agree that the entire Grovyle stage is pretty garbage, and Sceptile's ice weakness and leaf blade's low base power holds Sceptile back even against opponents that are weak to grass. I just read these posts and thought some of them are in fact too hard on Sceptile. For the record, people should probably test a Pokemon themselves before criticizing someone else for not testing enough.
When you post that thread, the tiers are still not finalized, right? I'm kind of busy right now but have more stuff I want to test in 3 or 4 months.