MVP-Shapley: Feature-based Modeling for Evaluating the Most Valuable Player in Basketball

TL;DR

MVP-Shapley addresses the challenge of explainable MVP evaluation from play-by-play data by framing MVP attribution as a Shapley-value problem. It trains a win-loss predictor on concatenated player features via LightGBM, then uses feature-level Shapley values to quantify each player's contribution to game outcomes, enabling single-game and multi-game MVP rankings. The approach includes causal refinements and fuzzification to align with expert voting, and demonstrates strong performance on NBA data and the Dunk City Dynasty esports dataset, with a scalable online deployment framework. The work advances interpretable, data-driven MVP evaluation for both traditional sports analytics and esports contexts, offering practical impact for ranking, coaching insight, and industry deployment.

Abstract

The burgeoning growth of the esports and multiplayer online gaming community has highlighted the critical importance of evaluating the Most Valuable Player (MVP). The establishment of an explainable and practical MVP evaluation method is very challenging. In our study, we specifically focus on play-by-play data, which records related events during the game, such as assists and points. We aim to address the challenges by introducing a new MVP evaluation framework, denoted as \oursys, which leverages Shapley values. This approach encompasses feature processing, win-loss model training, Shapley value allocation, and MVP ranking determination based on players' contributions. Additionally, we optimize our algorithm to align with expert voting results from the perspective of causality. Finally, we substantiated the efficacy of our method through validation using the NBA dataset and the Dunk City Dynasty dataset and implemented online deployment in the industry.

Figures (8)

• Figure 1: There are various player statistics in play-by-play data; the challenge is determining how to utilize these data to evaluate the MVP.
• Figure 2: Cause and effect graph.
• Figure 3: The dashed horizontal line represents the mean, the red horizontal line represents the median, and the orange triangles are outliers.
• Figure 4: Mean-variance plot of the evaluation metrics to crowdsourced confidence.
• Figure 5: Online Service Deployment Framework.