This study investigates the predictive power of current, short-, and long-term expected valence in decision-making within an experience-based learning task. Across two experiments participants engaged in a gambling task where they had to balance short- and long-term outcomes to maximize gains. In Experiment 1 participants had to accept short-term losses to achieve long-term gains, while in Experiment 2 they had to omit short-term gains. Results from generalized mixed-effects models revealed that all three valence constructs (current, short-term, and long-term expected valence) were significant predictors of risky choices, with their influence modulated by the specific choice context. In a loss context participants relied more on short-term expectations, while in an omission context long-term expectations played a stronger role. These findings align with existing literature on the influence of emotional valence on decision-making and demonstrate the adaptability of the subjective valuation system across different choice scenarios. The study highlights the importance of considering multiple emotional self-report dimensions in decision-making processes.