Two betting strategies, the exact same synthetic season, the exact same "true" edge on every game. The only difference between the two simulation modes below is how the random numbers get assigned to events — and that alone is enough to flip who looks like the winner.
rng.next() for every bet it actually places — so the draw index it lands on depends on its own control flow (how many bets it has skipped so far), not on which game is being resolved.hash(seed, game_id, "outcome") — a pure function of the event, not of position in a draw sequence. Whether a strategy bets on game 7 or skips straight to game 12, game 7's outcome is always the same fixed value.