In a sequential game, using backward induction guarantees that players will always choose the strategy that maximizes their payoffs, regardless of their initial preferences.

Backward induction works by looking at the last decision point in a sequential game and assuming the player there will pick the action that gives them the highest payoff. Knowing this, the player before that point can anticipate the last player’s choice and choose the action that maximizes their own payoff, given that anticipation. By repeating this reasoning back to the first move, each player selects a strategy that is optimal given the future optimal choices, so no player can improve by deviating. This process relies on rationality and the assumption that each player wants to maximize their own payoff, not on any particular initial preference. For example, in a two‑move game where the second player chooses between 5 or 3, the first player knows the second will pick 5, so the first chooses the action that gives them a payoff of 8 rather than 6, thus following backward induction.