Double ranking selection - distances and standard deviation

I have a matrix x (250x10 double), this are normalized prices. The columns are 10 assets and the rows represent the time The first 5 rows (of the 250 rows) looks as follow:

-0.730  -0.859  -1.490  0.505  1.546  1.376  -0.343  2.075  2.349  1.104
-0.510  -0.830  -1.346  0.392  1.510  1.135  -0.343  2.301  2.267  1.487
-0.344  -0.844  -1.225  0.335  1.528  0.813  -0.496  2.326  2.145  1.487
-0.694  -0.931  -1.153  0.221  1.473  0.813  -0.343  2.175  1.818  1.334
-0.142  -0.888  -1.104  0.221  1.510  1.054  -0.232  2.175  1.370  1.334

I want to calculate the distance for each row between all possible columns (assets). So I created a matrix Qdist (10x10) for all possible pairs. After that I create to matrices where I determine the smallest pair for each column(asset) [m] and the related pair column(asset) [n]. In the last step I rank the the pairs from 1 to 10, these ranks I want to use in the next step.

[n2]=size(x,2);

Qdist=zeros(n2,n2);

for i=1:n2
    for j=1:i-1
        Qdist(i,j)=sum((x(:,i)-x(:,j)).^2);
        Qdist(j,i)=Qdist(i,j);
    end
    Qdist(i,i)=nan;
end

[m,n]=min(Qdist);
pairs(:,1)=n';
[~,r]=sort(m);
[~,xRanks]=sort(r); %rank all stocks

After ranking the pairs from 1 to 10,I want to select only the 8 best pairs (Q1). With these 8 pairs I want to determine the standard deviation of each column of matrix x2 (just the prices, and not normalized). So all pairs have two different standard deviations. This means that I get 16 standard deviations (Q2).

Matrix x2 looks as follow (first 5 rows)

11.41  5.2  1.12  4.12  4.68  1.61  8.23  6.06  9.16  2.93
11.53  5.22  1.18  4  4.66  1.58  8.23  6.15  9.14  3.03
11.62  5.21  1.23  3.94  4.67  1.54  8.12  6.16  9.11  3.03
11.43  5.15  1.26  3.82  4.64  1.54  8.23  6.1  9.03  2.99
11.73  5.18  1.28  3.82  4.66  1.57  8.31  6.1  8.92  2.99

Next I want to select 4 pairs that is ranked at least in the top 4 highest standard deviations (Q3). So this means that I have to rank twice. (first on distance, second in st. dev.)

What are you recommendations for solving Q1,Q2,Q3?

I already came this far (see code), but the last part seems to be toughest part.

I hope it is clear to you all and I am eager to hear from you!

[Julian]

Although this relates to the first part of your problem, which I believe you have solved, the old FEX contribution distance.m is handy in this context....