This paper addresses the problem of attributing portfolio performance to individual investment features, such as active strategies or factors, plus a baseline component. It highlights that traditional attribution methods can yield inconsistent results, leading to ambiguity in performance evaluation. To resolve this, the authors propose using the Shapley value from cooperative game theory, which provides a principled way to allocate performance fairly among features by considering all possible combinations and interactions. This method ensures that attributions are unique and satisfy desirable properties like efficiency and symmetry, making it a robust alternative to ad-hoc approaches.

The research discusses practical implementation by detailing both exact computational techniques for scenarios with a limited number of features and approximate methods for more complex portfolios where exact calculation is computationally intensive. By applying Shapley attribution, the paper demonstrates how investors can gain clearer insights into the contributions of each feature, enhancing decision-making and performance analysis. This approach bridges theoretical game theory with quantitative finance, offering a standardized framework that can improve transparency and accountability in investment processes.