R, 106 96 bytes

function(n,k,l)n*mean(matrix(apply(expand.grid(rep(list(1:k),w<-n+abs(l))),1,sort,l>0),w)[1:n,])

Try it online!

Credit to Dominic van Essen for the l>0 for the descending argument to sort, and for 9 bytes golfed down.

function(n,k,l,w=n+abs(l)){
D <- expand.grid(rep(list(1:k),w))	# data.frame of all possible die rolls, across rows
S <-apply(D,1,sort,l>0)			# sort the rows of D in descending order if l <= 0 and ascending otherwise. This results in columns of dice rolls.
M <- matrix(S,w)			# make S a Matrix of w rows. This helps the edge case n = 1.
H <- head(M,n)				# take the first n rows of M
mean(colSums(H))			# take the mean of the dice rolls, and return
}

R, 106 96 88 bytes

function(n,k,l)n*mean(apply(expand.grid(rep(list(NA,1:k),n+abs(l))),1,sort,l>0,T)[1:n,])

Try it online!

Credit to Dominic van Essen for the l>0 for the descending argument to sort, and for golfing down a lot of other bytes!

R, 106 bytes

function(n,k,l,w=n+abs(l))mean(colSums(head(matrix(apply(expand.grid(rep(list(1:k),w)),1,sort,l>0),w),n)))

Try it online!

Credit to Dominic van Essen for the l>0 for the descending argument to sort.

function(n,k,l,w=n+abs(l)){
D <- expand.grid(rep(list(1:k),w))	# data.frame of all possible die rolls, across rows
S <-apply(D,1,sort,l>0)			# sort the rows of D in descending order if l <= 0 and ascending otherwise. This results in columns of dice rolls.
M <- matrix(S,w)			# make S a Matrix of w rows. This helps the edge case n = 1.
H <- head(M,n)				# take the first n rows of M
mean(colSums(H))			# take the mean of the dice rolls, and return
}

R, 106 96 bytes

function(n,k,l)n*mean(matrix(apply(expand.grid(rep(list(1:k),w<-n+abs(l))),1,sort,l>0),w)[1:n,])

Try it online!

Credit to Dominic van Essen for the l>0 for the descending argument to sort, and for 9 bytes golfed down.

function(n,k,l,w=n+abs(l)){
D <- expand.grid(rep(list(1:k),w))	# data.frame of all possible die rolls, across rows
S <-apply(D,1,sort,l>0)			# sort the rows of D in descending order if l <= 0 and ascending otherwise. This results in columns of dice rolls.
M <- matrix(S,w)			# make S a Matrix of w rows. This helps the edge case n = 1.
H <- head(M,n)				# take the first n rows of M
mean(colSums(H))			# take the mean of the dice rolls, and return
}