Journal of Sports Analytics - Volume 7, issue 4

Authors: Gollub, Jacob

Article Type: Research Article

Abstract: Many research papers on tennis match prediction use a hierarchical Markov Model. To predict match outcomes, this model requires input parameters for each player’s serving ability. While these parameters are often computed directly from each player’s historical percentages of points won on serve and return, doing so fails to address bias due to limited sample size and differences in strength of schedule. In this paper, we explore a handful of novel approaches to forecasting serve performance that specifically address these limitations. By applying an Efron-Morris estimator, we provide a means to robustly forecast outcomes when players have limited match data …over the past year. Next, through tracking expected serve and return performance in past matches, we account for strength of schedule across all points in a player’s match history. Finally, we demonstrate a new way to synthesize historical serve data with the predictive power of Elo ratings. When forecasting serve performance across 7,622 ATP tour-level matches from 2014-2016, all three of these proposed methods outperformed Barnett and Clarke’s standard approach. Show more

Keywords: Match prediction, point-based models, Efron-Morris estimator, Elo ratings

DOI: 10.3233/JSA-200345

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 223-233, 2021

Authors: Gift, Paul

Article Type: Research Article

Abstract: This paper investigates the impact of changes in judging criteria on 10-8 scores in Zuffa-owned mixed martial arts (MMA) promotions. Utilizing a differences-in-differences framework, the 2017 liberalization of 10-8 scoring criteria in the Unified Rules of MMA is examined across various judge groups. Findings suggest that traveling judges and Nevada judges – those most likely to be at the forefront of the regulatory evolution of the sport – had already liberalized their 10-8 scoring one year prior to the effective date of the new criteria. Other judges appear to have effectively implemented the new criteria since January 2017 with 10-8 …probabilities on par with traveling and Nevada judges. The effect of an earlier change in judging criteria is also examined in Nevada. Results suggest the numerous and distributed regulatory agencies involved in the sport of MMA were effective in the implementation of new policies for scoring rounds. Show more

Keywords: Judging, mixed martial arts (MMA), policy effectiveness, regulation, Ultimate Fighting Championship (UFC), Zuffa

DOI: 10.3233/JSA-200478

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 235-246, 2021

Authors: Chatzakis, Prokopios | Paradisis, Giorgos | Chryssanthopoulos, Costas | Zacharogiannis, Elias

Article Type: Research Article

Abstract: The purpose of the present study was to examine the effect of sex and performance standard on pacing profiles in a 24 h ultra-marathon race. Performance data of 283 participants (237 men and 46 women) from the last decade’s versions (2011 until 2020, with the exception of the 2017 version) of the International Ultramarathon Festival held in Athens-Hellinikon, Greece, were analyzed and pacing profiles were evaluated based on performance standard and sex. Relative speed for every hour and % distance covered in 6 h and 12 h segments and coefficient of variation (CV) were calculated. Mean distance ran was 159.99±36.04 km. Runners followed …a reverse J-shaped race pace (p  < 0.001). Sex did not seem to interact with pacing (p  > 0.05 in every case), while performance standard interacted significantly with pacing (p  < 0.001). CV was negatively correlated with total distance covered and total running time (–0.761, p  < 0.001 and –0.753, p  < 0.001, respectively). In conclusion, the overall pacing profile adopted by runners in a 24 h ultra-marathon race was a reverse J-shaped model, with better runners following a more even pacing than slower runners, with lower velocity variability. Show more

Keywords: Ultra-endurance, distance running, effort distribution, performance, sex

DOI: 10.3233/JSA-200496

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 247-253, 2021

Authors: Gao, Zijian | Kowalczyk, Amanda

Article Type: Research Article

Abstract: Tennis is a popular sport worldwide, boasting millions of fans and numerous national and international tournaments. Like many sports, tennis has benefitted from the popularity of rigorous record-keeping of game and player information, as well as the growth of machine learning methods for use in sports analytics. Of particular interest to bettors and betting companies alike is potential use of sports records to predict tennis match outcomes prior to match start. We compiled, cleaned, and used the largest database of tennis match information to date to predict match outcome using fairly simple machine learning methods. Using such methods allows for …rapid fit and prediction times to readily incorporate new data and make real-time predictions. We were able to predict match outcomes with upwards of 80%accuracy, much greater than predictions using betting odds alone, and identify serve strength as a key predictor of match outcome. By combining prediction accuracies from three models, we were able to nearly recreate a probability distribution based on average betting odds from betting companies, which indicates that betting companies are using similar information to assign odds to matches. These results demonstrate the capability of relatively simple machine learning models to quite accurately predict tennis match outcomes. Show more

Keywords: Tennis, machine learning, sports analytics, random forest, feature selection, betting

DOI: 10.3233/JSA-200515

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 255-262, 2021

Authors: Rocha da Silva, João Vítor | Rodrigues, Paulo Canas

Article Type: Research Article

Abstract: The NBA (National Basketball Association) is going through a transition process in its way of practice, planning, and comprehension of the game. With the exponential growth of the data that has been collected, detailed statistical analyses have been conducted for each part of the game. This has been overwhelmingly exploited in a way never seen before, especially when dealing with the three-point shot. In this paper, we are interested in characterizing NBA’s gameplay over time to identify trends and success factors. In particular, this study aims: (i) to identify which factors were crucial for teams’ regular season success in …the past and understand the factors that are more relevant to succeed in the present day; and (ii) to group seasons and regular season winning teams into clusters of common characteristics and gameplay behavior. Historical events and trends help us to understand how teams were successful in past regular seasons, how they played, and how their style of play has changed. Leading to a better comprehension of the game. The game-related statistics of the NBA’s regular seasons, from 1979-80 to 2018-19, were analyzed using principal component analysis, cluster analysis, and LASSO regression. It is possible to identify three main Eras that we define as the Classic Era of the NBA (1980–1994), the Transitional Era of the NBA (1995–2013), and the Modern Era of the NBA (since 2013). As the results of this study make a historic analysis of the NBA, indicating the three eras of NBA regular seasons since the introduction of the three-point line, their playing styles, and their respective factors for success, this present research may be the base study that will help researchers better investigate the NBA, its past, present, and future. Show more

Keywords: Basketball, data science, K-means clustering, LASSO regression, national basketball association, principal component analysis, sports analytics, statistics, unsupervised machine learning.

DOI: 10.3233/JSA-200525

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 263-275, 2021

Authors: Cohan, Alexander | Schuster, Jake | Fernandez, Jose

Article Type: Research Article

Abstract: Predicting athlete injury risk has been a holy grail in sports medicine with little progress to date due to a variety of factors such as small sample sizes, significantly imbalanced data, and inadequate statistical approaches. Data modeling which does not account for multiple interactions across factors can be misleading. We address the small sample size by collecting longitudinal data of NBA player injuries using publicly available data sources and develop a state of the art deep learning model, METIC, to predict future injuries based on past injuries, game activity, and player statistics. We evaluate model performance using metrics appropriate for …imbalanced data and find that METIC performs significantly better than other traditional machine learning approaches. METIC uses feature learning to create interactive features which become meaningful in combination with each other. METIC can be used by practitioners and front offices to improve athlete management and reduce injury incidence, potentially saving sports teams millions in revenue due to reduced athlete injuries. Show more

Keywords: Artificial intelligence, modelling, analytics, injuries, basketball

DOI: 10.3233/JSA-200529

Citation: Journal of Sports Analytics, vol. 7, no. 4, pp. 277-289, 2021