Date of Award

2025

Document Type

Thesis

Degree Name

Bachelors

Department

Natural Sciences

First Advisor

Skripnikov, Andrey

Area of Concentration

Statistics

Abstract

We explore the within-game variables that can be used to project the typical number of goals a National Hockey League (NHL) team scores in a game. In particular, we used Poisson-type count-response models to model expected goals for a game given other variables such as shots, faceoffs won, penalty minutes, etc. Subsequently, we analyzed the teams whose actual scoring production over- and underperformed our model’s expected goals projection in the regular season, and investigated whether it could be used for predicting their performance in the playoffs. The data used to conduct this research are regular season and postseason game logs for every NHL team over the past 9 seasons (2016-2024). The data was scraped from hockey-reference.com.

Download

Share

COinS