It has been demonstrated that ML can be used to produce superhuman SSBM agents, e.g. the Phillip project. At some point, I would like to try training my own ML agent with some tweaks on the Phillip methodology. I am a hobbyist programmer and have very little ML experience. I will be relying almost entirely on existing resources and LLMs to guide me through the process.

After resolution, I will conduct a best-of-five set (under the competitive ruleset at the time) against the strongest available Phillip agent at the time (per the developer's evaluation).

Resolves to YES if my agent wins the Bo5 set.

Resolves to NO otherwise (including if I am unable to produce a working agent at all).