how to search and keep all the possible combinations with a certain accuracy

I am trying to do some simple combination, I have a Matrix of X consists of 2000 variables and a Y I want to each time pick 4 of the variables from X and make a regression and keep only those sets that have higher R than 0.9 . I found another solution based on 3 matrices, what want is one matrix and I tried to amend the code but somewhat I did not get it right

tic
X1 = rand(40,2000);
Y  = rand(40,1);
in = ones(40,1);
R = zeros(size(:,1),1);
for c = 1:size(:,1)
    X    = [in X1(:,i)];
    yhat = X*(X\Y);
    ybar = mean(Y);
    ssr  = norm(yhat - ybar)^2;  
    sst  = norm(Y - ybar)^2; 
    R(c) = ssr/sst;
end
toc