Predicting NFL Team Performance with ELO Rating, Home-Field Advantage, and QB Rating

Preregistration of analyses

Analysis #1

Question: How well does this ELO rating predict regular season / postseason game wins?

Analysis 1:

We will conduct a regression analysis to test whether ELO ratings are a significant predictor of game wins, for our second research question. To analyze the data, we will pick some teams as samples, using regression to analyze the relationship between ELO ratings and game wins. Some teams in mind for this analysis are the New York Giants and New York Jets. Additionally, by using the `elo_prob` variable, we can deduce how well this variable predicted actual wins and whether the variable got more predictive over time. 

Overall, the analysis will show us whether the data itself is a good predictor of team success in the NFL. Oftentimes in the media today, tv stations and internet sources will provide probabilities of an NFL team winning a game. Using ELO ratings, we can see whether this variable predicts success well and whether or not it can be used to generate win probabilities. In the analysis, we will utilize linear regression and charts to visualize the variable’s predicting validity.

Analysis #2

Question: Does having the home field really give a team an advantage? Is this advantage different or more pronounced in the playoffs?

Analysis 2:

We will conduct a correlation analysis to identify any significant changes in team performance on home field in the regular season / postseason.

We will also conduct a hypothesis test. After selecting some teams such as the New York Giants and New York Jets, we will test hypotheses for each one. The null hypothesis is the probability of winning in the home field equals the probability of winning in the away fields, while the alternative hypothesis is that the probability of winning in the home field does not equal the probability of winning in the away fields. We will then calculate the p-value to test the significance and interpret it in the context. Confidence intervals will be constructed if needed.

To improve overall performance of this model, the group may consider combining it with the model above. In the NFL, playing at the home field or the away field is not the only consideration in determining the outcome of the game. Using strength of schedule, ELO ratings, and other metrics may also be important in determining whether a team will win or lose. Overall, this model will help us determine if playing in one’s own home field truly has an impact on a team’s success or not.