Best-of + Worst-of = Call1 + Call2
The right hand side is independent of correlation (and you can check it in your model).
Therefore if Best-of is short correlation, worst-of must be long correlation.
Increasing correlation makes the two assets more similar and therefore makes the best-of more like a vanilla. This is why a best-of is short correlation.
Intuitively, they should both be short correlation, that is the less correlated the assets are the higher the value of the worst of/best of option.
The best of option payoff is sandwiched by an exchange option payoff (plus other vanilla forward/option payoffs on single stock, insensitive to correlation):
$$ X_T -K + (Y_T-X_T)^+ \leq \max(X_T - K ,Y_T - K,0) \...
Since the correlation matrix is symetric, if you move the term (i,j), you have to do it for the term (j,i) as well
Of course -> the correlation of an asset with itself is equal to 1... so it should not change
You apply a downward shock (1 to 0.99) and you use the formula of finite differences