Introduction


The article describes an approach to ranking NLF teams over time. The method used was to calculate the ELO ranking for every game in NFL history.

library(devtools)
## Loading required package: usethis
library(RCurl)
library(plyr)

url_538 <- getURL("https://projects.fivethirtyeight.com/nfl-api/nfl_elo_latest.csv")
full_df <-data.frame(read.csv(text=url_538, header=T))


elo_df <-subset(full_df,team1=="DEN",select=c(date,season,team1,team2,elo1_pre,elo2_pre,score1,score2))
head(elo_df)

The following graph shows the pre game elo probablities for the DEN home games during the 2020 season

plot(x=elo_df$elo1_pre, type="o")



Findings and Recommendation


The pre game ELO data and the resulting scores could be used to evaluate the effectivness of ELO as a predictor of team performance.