Evaluating the Efficacy of Talent Identification and Development in the National Hockey League Entry Draft

Objective: The overarching objective of this dissertation was to evaluate the efficacy of talent selection and development in the National Hockey League entry draft, as well as to inform talent selection and draft decisions to improve accuracy. This objective was carried out by addressing three main research questions: (a) How accurately can NHL decision makers predict future performance through the NHL draft (Chapter 2)? (b) How do the variables that predict draft status fare in predicting subsequent NHL success (Chapter 3)? (c) Do sunk cost effects exist in the NHL (Chapter 4)? Methods: To address the first research question in Chapter 2, draft data (round and pick number) and subsequent NHL performance data were collected for the draft classes of 2007-2014. Two novel measures of offensive and defensive NHL performance were constructed using Exploratory Factor Analysis (EFA), then retrospectively compared such measures across all seven rounds of the draft using Kruskal-Wallis and Mann-Whitney U tests. In Chapter 3, pre-draft performance metrics, anthropometric measures, and subjective text-mined player attributes were collected. The relationship between such predictor variables and NHL order was studied using linear regression (part 1), before using Zero-Inflated Negative Binomial (ZINB) regression to study the relationship between the same independent variables and future NHL success (defined by games played). Lastly, Chapter 5 tested for sunk cost effects by using a hierarchical linear regression model which tested the relationship between draft order and NHL Time On Ice (TOI), while controlling for injury, trades, and on-ice performance. Results: Findings from Chapter 2 showed NHL decision makers' ability to project future performance declines markedly after the first two rounds in forwards, and after the first round in defensemen. In Chapter 3, results indicated that forwards' penalty minutes, height, age, and lack of physical strength were overvalued in the draft, while plus/minus, board battles, backchecking ability, and hockey sense were undervalued. For defensemen, NHL teams overvalued CSB rankings, and undervalued playmaking ability, lack of physical strength, and work ethic/leadership. Last, findings in Chapter 4 found evidence of sunk cost effects, where players selected in the first round received significantly more playing time than their counterparts drafted later, even after controlling for confounding variables. Conclusion: This dissertation uncovered substantial selection inaccuracies in the NHL entry draft, and identified a plethora of inefficiencies that drive these inaccuracies. Moreover, it highlighted that late draftees' development is compromised due to the sunk cost fallacy. As a result, this dissertation had significant implications for both research and practice. More specifically, it highlights avenues for future research to study draft related decision-making errors in more depth, and offers a range of suggestions to NHL personnel on improving their talent selection and development strategies