Why should the baseline be "randomness?"
Suppose the following. Two cloned teams. 14 games.
Team A is managed more or less randomly. A coin flip is used to determine which bench player to PH with, when such should occur, and what pitching changes will be made.
Team B is managed tactically. Meaning decisions are made about the current game in context of not only that game, but the upcoming one. PHs and relief appearances are used to their maximum effect.
Nobody seriously thinks Team A and Team B will split the series, or that the derivation in outcomes is just "luck."
Ergo, what we all already know. Baseball decisions matter in discrete games. They also matter over a series of games.
If you want to know why a team under or over performed in a discrete set of games, you first have to actually observe those games and see what patterns there are. There's really just no getting around that. If a particular series is won or lost on a series of bloop hits - that's what it is. If the teams are grossly outmatched in talent and that reflects normally - that's what it is. If the losing team makes sub-optimal decisions that impact the games - that's what it is.