Matching unequal data frames on a unique identifier - r

Question

So I have all this data from the EPL and I'm currently trying to create a column for a teams "form" based on their last five games. A win counts at 1 point, a draw as .5, and a loss zero. I have a loop to do this for a single teams at a time but when I try to create one to merge them all together I can not get it to work for some reason. My data comes from: http://www.football-data.co.uk/englandm.php
For the purpose for making it more simple I will use only the data from the 2013/2014 season for the Premier League.
I import the data from the excel sheet and labeled it PL_20
library(RCurl)
URL <- "www.football-data.co.uk/mmz4281/1415/E0.csv"
x <- getURL(URL, ssl.verifypeer = FALSE)
PL_20 <- read.csv(textConnection(x))
# cleaning data, getting rid of the betting odds
data <- PL_20[,-c(1,10,11,24:65)]
#Get dates all in same date format
data$Date<-as.Date(data$Date, guess_formats(data$Date, "dmy"))
#sorting (used for all the seasons)
sorted <- data[order(data$Date,decreasing=TRUE),]
sorted$index <- seq(1:nrow(sorted))
# making a column for form
teams<-as.matrix(unique(data$HomeTeam))
test<-sorted
z<-1 # controls which team the form is being found for. Ideally I would have this cycle
# through all of the teams
current<-subset(sorted, HomeTeam==as.character(teams[z]) | AwayTeam==as.character(teams[z]))
current$h.form<-0
current$a.form<-0
current$recent<-0
for (i in 1:nrow(current)){
if((as.character(current[i,2])==as.character(teams[z]) && as.character(current[i,6])=="H") || (as.character(current[i,3])==as.character(teams[z]) && as.character(current[i,6])=="A")){
# current[i,7]<- "W"
current[i,24]<- 1
}else{
if((as.character(current[i,2])==as.character(teams[z]) && as.character(current[i,6])=="D") || (as.character(current[i,3])==as.character(teams[z]) && as.character(current[i,6])=="D"))
{
#current[i,7]<- "D"
current[i,24]<- .5
}else{
if((as.character(current[i,2])==as.character(teams[z]) && as.character(current[i,6])=="A") || (as.character(current[i,3])==as.character(teams[z]) && as.character(current[i,6])=="H"))
{
# current[i,7]<- "L"
current[i,24]<- 0
}
}
}
}
d<-0
for (d in 0:(nrow(current)-6))
{
if (as.character(current[nrow(current)-(5+d),2])==as.character(teams[z])){
current[(nrow(current)-(5+d)),22]<-as.numeric(sum(current[(nrow(current)-(4+d)):(nrow(current)-d),24]))
}else{
if(as.character(current[nrow(current)-(5+d),3])==as.character(teams[z]))
{
current[(nrow(current)-(5+d)),23]<-sum(current[(nrow(current)-(4+d)):(nrow(current)-d),24])
}
}
}
Now these ugly loops create a data frame called current which has three columns at the end: h.form, a.form, and recent. h.form is the form for the designated home team for that game and a.form is the form for the designated away team for that game. Recent is just the outcome of that game.
I would like to be able to combine all of the teams together so there is one observation per game and both h.form and a.form are populated with the correct values for their corresponding teams.
Your help is appreciated and if you have suggestions on how to clean up these loops that would be helpful as well.