Mark E. Glickman, JQAS Editor-in-Chief

The June 2016 issue (volume 12, issue 2) of the Journal of Quantitative Analysis in Sports (JQAS) consists of four articles with applications to basketball, hockey, baseball, and car racing.

“Estimating an NBA Player’s Impact on His Team’s Chances of Winning” by Sameer Deshpande and Shane Jensen is the Editor’s Choice article for this issue and available for free download for the next 12 months. The article develops an approach to evaluating player contributions in the NBA using a two-step process. They first construct a win-probability model for NBA games that estimates the probability of a game outcome as a function of time remaining in the game and the current score difference. Using the fitted probabilities from this model, they evaluate player contributions to the change in the fitted probabilities through Bayesian linear models that control for other players on the court and other important factors. They demonstrate their approach on seven years of NBA play-by-play data and construct player impact scores that permit assessments of players’ worth to their teams.

“Formula for Success: Multilevel Modeling of Formula One Driver and Constructor Performance” by Andrew Bell, James Smith, Clive Sabel, and Kelvyn Jones models points scored by drivers in Formula 1 races in a cross-classified multilevel model that partitions variance into team, team-year, and driver levels to measure driver ability conditional on team performance. The linear effects in the model are then allowed to vary by year, track type, and weather conditions using complex variance functions. Using 905 Formula 1 races between 1950 and 2014, the authors address the question of who is the best Formula 1 driver of all time.

“Improved Component Predictions of Batting and Pitching Measures” by Jim Albert decomposes standard measures of batting and pitching performance into subcomponents, and then considers specification of the measures through telescoping products of conditional probabilities. The subcomponents are modeled as multinomial distributions within a Bayesian framework. The approach is demonstrated on Major League Baseball data and applies the decomposition approach to batting averages, on-base percentage, and the Fielding Independent Pitching measure for pitching success.

Finally, “Beating the Market on NHL Hockey Games” by Samuel Buttrey develops a method for predicting the outcome of National Hockey League (NHL) games. The goal-production process is modeled by a pair of Poisson processes, producing goals at a rate that is assumed constant for each manpower situation over the course of a game. The author constructs a betting strategy using the probabilities predicted on 2013—2014 validation data with favorable results.

These articles are available to all members of the Statistics in Sports Section and on a subscription basis from the JQAS website. Prospective authors can find the journal’s aims, scope, and manuscript submission instructions on site, as well.