In the sliding window technique, we do not iterate through each possible subarray. This is why the algorithm is so efficient!

But we do iterate through some subarrays!

The question is: which subarrays do we iterate through?

At each iteration, the current window is [left, right]. We shrink the window by moving left, until the window is valid.

So we can say that at each iteration, the window is the largest possible valid window that ends at right. If we increased the size of the window (as right is fixed, we would have to move left), the window would no longer be valid.

So we can say that at each iteration, we create a valid window that is the largest possible valid window that ends at right.