[论文解读] DeepHoops: Evaluating Micro-Actions in Basketball Using Deep Feature Representations of Spatio-Temporal Data

Anthony Sicilia, Konstantinos Pelechrinis+1 more|arXiv (Cornell University)|Feb 21, 2019

Sports Analytics and Performance被引用 6

一句话总结

DeepHoops 提出了一种深度学习框架，利用NBA比赛中的时空追踪数据，实时估算预期得分，将进攻回合建模为序列分类问题，以评估挡拆、切入和传球等微观动作。该模型实现了校准良好的概率估计，并量化了单个动作带来的预期得分提升，结果显示助攻在超过72%的时间内提升了预期得分。

ABSTRACT

How much is an on-ball screen worth? How much is a backdoor cut away from the ball worth? Basketball is one of a number of sports which, within the past decade, have seen an explosion in quantitative metrics and methods for evaluating players and teams. However, it is still challenging to evaluate individual off-ball events in terms of how they contribute to the success of a possession. In this study, we develop an end-to-end deep learning architecture DeepHoops to process a unique dataset composed of spatio-temporal tracking data from NBA games in order to generate a running stream of predictions on the expected points to be scored as a possession progresses. We frame the problem as a multi-class sequence classification problem in which our model estimates probabilities of terminal actions taken by players (e.g. take field goal, turnover, foul etc.) at each moment of a possession based on a sequence of ball and player court locations preceding the said moment. Each of these terminal actions is associated with an expected point value, which is used to estimate the expected points to be scored. One of the challenges associated with this problem is the high imbalance in the action classes. To solve this problem, we parameterize a downsampling scheme for the training phase. We demonstrate that DeepHoops is well-calibrated, estimating accurately the probabilities of each terminal action and we further showcase the model's capability to evaluate individual actions (potentially off-ball) within a possession that are not captured by boxscore statistics.

研究动机与目标

• 开发一种评估传统记分卡统计数据未涵盖的篮球无球微观动作的方法。
• 利用实时时空数据建模进攻回合中预期得分的动态变化。
• 通过参数化下采样方案解决终端动作预测中的类别不平衡问题（例如，投篮、失误）。
• 通过测量预期得分的变化，量化特定微观动作（例如传球、挡拆）的价值。
• 支持教练和战略决策中的实时分析与“如果……会怎样”情景分析。

提出的方法

• DeepHoops 使用深度学习架构处理时间窗口 $W$ 内球员和球位置的 $T$ 个时间步的时空序列。
• 该模型在每个进攻回合的每个时刻预测终端动作（例如，投篮、失误、犯规）的多类别概率分布。
• 通过引入球员嵌入来考虑人员差异，实现对动作的上下文感知评估。
• 预期得分 $\epsilon_{\pi}(\tau)$ 通过终端动作概率及其对应得分值的加权和计算得出。
• 训练期间应用参数化下采样方案，以解决如犯规或失误等罕见事件的高类别不平衡问题。
• 该框架计算预期得分提升 $\Delta \epsilon_{\pi}(\alpha, \tau)$，即在时间 $\tau$ 执行微观动作 $\alpha$ 前后预期得分的差值。

实验结果

研究问题

• RQ1 如何从预期得分结果的角度，对挡拆、切入和传球等微观动作的影响进行定量评估？
• RQ2 深度学习模型能否利用时空追踪数据实时准确估算终端动作（例如，投篮、失误）的概率分布？
• RQ3 与普通传球相比，助攻在多大程度上提升了预期得分？该差异是否具有统计显著性？
• RQ4 该模型对罕见事件（如投篮犯规或失误）的概率估计校准程度如何？
• RQ5 DeepHoops 是否能够支持教练和实时比赛决策的“实时”或“如果……会怎样”情景分析？

主要发现

• DeepHoops 对终端动作的概率估计具有良好的校准性，表明其预测具有可靠的置信度。
• 超过72%的助攻提升了预期得分，表明其与战略价值高度一致。
• Kolmogorov-Smirnov 检验在 0.1% 显著性水平下拒绝了助攻与普通传球的预期得分提升分布相等的原假设。
• 该模型能成功识别出可能发生投篮犯规的时机——例如，当球员突破至禁区时——通过防守者靠近而提高犯规概率。
• 实时快照显示，DeepHoops 在关键微观动作（如挡拆与传球导致空位三分投篮）后检测到预期得分的提升。
• 该框架能有效区分高影响力与低影响力传球，助攻带来的预期得分提升显著高于非助攻传球。