| football {BayesDA} | R Documentation |
Data on football point spreads and game outcomes (north american football) for ten seasons, 1981, 1983-1986, 1988-1992, each season are 224 games and they are strung together. Only three first seasons are used in chapter one of book.
data(football)
A data frame with 2240 observations on the following 7 variables.
homefavoriteunderdogspreadfavorite.nameATL BUF CHI
CIN
CLE DAL DEN DET GB
HOU IND KC LAA LAN
MIA MIN NE NO NYG
NYJ PHA PHX PIT SD
SEA
SF TB WASunderdog.nameATL BUF CHI
CIN CLE
DAL DEN DET GB HOU IND
KC LAA
LAN MIA MIN NE NO NYG
NYJ PHA
PHX PIT SD SEA SF
TB WASweekFootball experts provide the point spread as a measure of the difference in ability between the two teams. For example, team A might be a 3.5 favourite to team B. The implication of this is that the proposition that team A, the favourite, defeats team B, the underdog, by 4 or more points, are considered a fair bet. In other words, the probability that A wins by more than 3.5 points is 0.5. If the point spread are an integer, then the implication is that team A is as likely to win by more points than the point spread as it is to win by fewer points than the point spread (or to loose). If the win is by exactly the point spread then neither side is paid off.
data(football) summary(football) names(football) # In chapter 1 only three first seasons are used: cap1 <- football[1:672, ]