This paper introduces a novel approach to asset allocation by incorporating time-varying probability weighting derived from option-implied distributions. Traditional forward-looking models typically assume risk aversion under Expected Utility Theory, overlooking investors’ behavioral biases in probability perception. By embedding a probability distortion mechanism into the pricing kernel – estimated from options data – we capture how investors overweight or underweight tail risks over time. Empirically, we apply this method to the S\&P 500 index and options market, demonstrating that the adjusted real-world return distributions significantly improve portfolio performance across Sharpe ratios, opportunity cost, and utility-based measures. The gains are especially pronounced during crisis periods, where conventional models often fail. Robustness checks across alternative weighting functions, optimization schemes, and dynamic settings affirm the strategy’s stability. These findings underscore the significance of incorporating the time-varying probability weighting feature when utilizing option information for asset allocation.