Applying technical trading rule to all combination of inputs - r

Question

I have a Dataframe with columns containing different time series of data. I need to insert these columns into the function below automatically to find the best combination (highest return):
returns <- as.data.frame(rep(0, 4261)) #4261 because that's the length of n1.lc
returns$n2_5_10 <- rep(0, nrow(returns))
MSVrule <- function(n1, n2, hold){
for(i in 13:nrow(n1.lc)){
if (n1[i] > n2[i] & n1[i-1] < n2[i-1]) {
returns$n2_5_10[i] <- dt.lc$Settle[i - hold] - dt.lc$Settle[i]
} else {
if (n1[i] < n2[i] & n1[i-1] > n2[i-1])
{returns$n2_5_10[i] <- dt.lc$Settle[i] - dt.lc$Settle[i - hold]
}
else{
NULL
}
}
}
}
MSVrule(n1.lc$N1_2_5, n2.lc$n2_2_10, 5)
This function doesn't work, leaving returns$n2_5_10[i] 0 However, when I specifiy the vectors in the function it works:
hold <- 5
for(i in 13:nrow(n1.lc)){
if (n1.lc$N1_2_5[i] > n2.lc$n2_2_10[i] & n1.lc$N1_2_5[i-1] < n2.lc$n2_2_10[i-1]) {
returns$n2_5_10[i] <- (dt.lc$Settle[i - hold] - dt.lc$Settle[i]) / dt.lc$Settle[i]
} else {
if (n1.lc$N1_2_5[i] < n2.lc$n2_2_10[i] & n1.lc$N1_2_5[i-1] > n2.lc$n2_2_10[i-1])
{returns$n2_5_10[i] <- (dt.lc$Settle[i] - dt.lc$Settle[i - hold]) / dt.lc$Settle[i - hold]
}
else{
NULL
}
}
}
The next step would be to automatically apply the function to other combinations of vectors from the n1.lc Dataframe. But I need the function to work first.

Answer 1

Because you use a function with no return() attempting to modify objects outside its scope, nothing external will be changed. However, you can use <<- operator to modify external objects to the function, namely the n2_5_10 column in returns:
Below demonstrates with random generated data and assigns the column in returns with 500 or 5 as you do not post other needed, used objects (i.e., dt.lc). Adjust to actual data objects:
n1.lc <- data.frame(
N1_2_5 = runif(50),
n2_2_10 = runif(50)
)
returns <- data.frame(n2_5_10 = rep(0, nrow(n1.lc)))
MSVrule <- function(n1, n2, hold){
for(i in 13:nrow(n1.lc)){
if (n1[i] > n2[i] & n1[i-1] < n2[i-1]) {
returns$n2_5_10[i] <<- 500
} else {
if (n1[i] < n2[i] & n1[i-1] > n2[i-1])
{returns$n2_5_10[i] <<- 5
}
else{
NULL
}
}
}
}
MSVrule(n1.lc$N1_2_5, n1.lc$n2_2_10, 5)
However, consider using return where you create/update/output returns inside function. Then on outside, assign a new dataframe as the function's output (no need for else { NULL } clause):
MSVrule <- function(n1, n2, hold){
returns <- data.frame(n2_5_10 = rep(0, nrow(n1.lc)))
for(i in 13:nrow(n1.lc)){
if (n1[i] > n2[i] & n1[i-1] < n2[i-1]) {
returns$n2_5_10[i] <- 500
} else {
if (n1[i] < n2[i] & n1[i-1] > n2[i-1])
returns$n2_5_10[i] <- 5
}
}
return(returns)
}
newdf <- MSVrule(n1.lc$N1_2_5, n1.lc$n2_2_10, 5)
# BOTH ABOVE RESULT IN EQUIVALENT OUTCOMES
all.equal(returns, newdf)
# TRUE
identical(returns, newdf)
# TRUE