Spring 2024 Archives

Evaluating the Impact of Special Teams on Winning

This paper evaluates the per-play impact of special teams on the winning percentage of NFL teams compared to offense and defense. Employing data from the 1999-2022 NFL seasons in the nflfastR database, we developed a new calculation for Expected Points Added (EPA) to accurately assess the contribution of special teams plays, including punts, kickoffs, and field goals, independent of coaching decisions. This methodology allowed us to create our Special Teams Performance Index (STPI), a composite metric that consolidates the performance of all special teams units on a team. We then used STPI in combination with offensive and defensive ratings from ProFootballReference in a regression model to evaluate the importance of special teams on winning percentage. Finally, since there are fewer plays on special teams than on offense or defense in a game, we adjusted the resulting numbers to find the impact on a per-play basis.…Read More

Exploring the Evolution of the NFL Draft Pick Trade Market Over Time

Building on the work of Cade Massey and Richard Thaler in 2005, this paper investigates changes and trends in the market for National Football League draft picks over the past 40 years. This period has seen significant advancements in the way NFL franchises approach the draft, including the Jimmy Johnson trade value chart, first employed by the Dallas Cowboys in the 1990s. Still, many (including Massey and Thaler) argue that teams continue to significantly overvalue early draft picks. This paper uses a Weibull distribution to model the market value of picks based on their position and year. Looking at each ten-year window, this study generates “rolling” variables to represent trade frequency, market consensus, discount rate, and the rate at which pick value decreases as pick number increases. The graphs of these four variables inform the division of the 40-year period into distinct phases. The findings suggest some maturation of the trade market over time, characterized by more trading and lower discount rates. However, the value decay rate has fluctuated, without a clear downward trend. This paper highlights the complex interplay between innovation, tradition, strategy, and psychological biases in the NFL draft.…Read More

High Value Football Transfers: A Win or Loss? Correlation of High Value Football Transfers, with their Subsequent Performances

The study is based on the correlation of high value football transfers, and the change in market value of those transfers for their new clubs. Analyzing pre covid data, from seasons 2017/18 to 19/20, a comparison has been made for the change in market value of the top 10 valued transfers, for goalkeepers, defenders, midfielders, attackers over the course of 2 years from their transfer. There has been minimal previous research on this, however, there has been research on why football transfers do not work out well. A variety of reasons such as not fitting in the team, locational preferences, problems with the coach etc. it has always been of great interest as to why players who are bought for such amounts, tend to not live up to their potential. A club makes this incredible investment, so that the player plays well in the club and improves it, whereas it seems to be the opposite. Transfer fee seems to have an unexpected negative correlation with player performances, which has been verified by the results in this experiment also. We can also see that players with high value fees who transfer to Spanish clubs seem to have the greatest dip in performances.…Read More

Analysis of Success Probabilities in Field Hockey with Machine Learning

The main goal of this project is to use modern machine-learning techniques to analyze field hockey player performance. Field hockey is a relatively heretofore unexplored application, and there would be a great benefit in building a model to predict players’ ability to score under different conditions. We have scraped public data from Northeastern’s 2023 field hockey season, and implemented a prototype mixed effects logistic regression model with both glmer [2] and RStan [6]. The model investigates effects that could impact players’ goal-scoring probability, including match location, score difference between the competing teams, and situational effects such as scoring off a penalty corner or not. The model also estimates random effects capturing individual players’ likelihood to score a goal. By interacting player effects with whether a shot came after a penalty corner, the model supports an expectation from the Northeastern coaching staff that some players perform worse when shooting after penalty corners due to game strategy.…Read More

The Sharpe Ratio and Hitter Evaluation: A New Application of Modern Portfolio Theory

Modern Portfolio Theory aims to optimize risk-adjusted returns by identifying assets and creating portfolios with the highest Sharpe Ratio. Generally, there are two strategic approaches to optimizing risk-adjusted returns: maximizing returns or minimizing volatility. In traditional financial literature, it is generally understood that forecasting the future returns of an asset by using its historical returns as a proxy yields low correlation and limited accuracy. However, forecasting the future volatility of an asset is a much more precise science due to the autocorrelation of its squared returns resulting in volatility clusters. In this paper, I will draw comparisons between the ways returns and volatility are measured in financial markets and the ways they can be applied in baseball analytics. Furthermore, I will provide a framework for hitter evaluation by contextualizing the historical difficulty of predicting financial returns accurately, while capitalizing on the predictive nature of volatility.…Read More

Understanding the Relationship Between Mental Health Literacy and Grit in a Sample of High School Female Athletes

A survey was conducted with 48 female participants (athletes n = 27, nonathletes n = 21) from a rural high school in Utah to determine the correlation between mental health literacy and grit. Each participant completed the modified Multicomponent Mental Health Literacy Measure (MMHLM), shortened Grit-S scale, challenge and commitment constructs from the Mental Toughness Questionnaire 48-item (MTQ48), and additional questions about their views of elite athletes and mental health.…Read More

Using a Novel Metric of Expected Points Above Average (EPAA) Versus Salary to Assess 2023 National Football League Kicker Value

In the National Football League (NFL), team salary caps mean money spent on players on one position reduces the ability to spend resources on others. Salary cap hits for kickers in 2023 ranged from under $1M to almost $6M, with cash salaries exhibiting an even wider range (<$1M to >$9M). Kickers have been primarily judged on the overall accuracy of their kicks within specific kick ranges. However, such an approach does not include factors such as wind, humidity, temperature, and other variables that could affect the probability of a successful kick.…Read More

A Holistic Examination of Streakiness and Consistency in Major League Baseball

Streakiness and baseball go hand in hand, but accurately measuring streakiness and consistency in sports is difficult. While studying hitting streaks is an old idea, relatively few works have examined streaks for hitters at the pitch outcome granularity, or for pitchers more generally. Furthermore, little is understood about how streaks correlate with more traditional player outcomes.…Read More

SET: Spatial Edge Technique – A Framework to Evaluate Edge Setters

Defensive plays on the football field require the effort of 11 players. However, defensive statistics only credit one or two defenders for the play’s end result (sack, tackle, interception, etc.). Our project focused on the off-ball contributions of the oft-ignored nine or ten others. In particular, we focused on a vital skill to stop the run — setting the edge.…Read More

Analyzing Force in Tackling with the Creation of the Force Over Expected Metric

Given the access to valuable play-by-play data through the 2024 Big Data Bowl, we decided to investigate the force generated by tackles. We hypothesized that the force of the tackle is meaningful in tackling success and sought to quantify it accurately. By modeling this metric, which we will refer to as “force generated at the spot of the tackle” against an array of potentially significant variables, we wanted to create an expected force metric that could be compared to the actual force generated on a specific play. From experimentation and analysis, we hoped to generate actionable insights and inspire further research into the relationship between force and tackling in the NFL.…Read More