To be honest though, doesn't the SV_Position semantic go from [0, screenWidth] x [0, screenHeight] from the top left corner? If so, if there's a pixel at (10,10), the bottom one would be (10,11) and top (10,9), wouldn't it?
I know that OpenGL is flipped upside down (like TGA) where the bottom left is (0,0), I'm not sure about D3D though. In reality, it doesn't matter if you call it top/bottom head/toe, etc.. What matters is that the 'greater' value has the 'lesser' value subtracted - regardless of the orientation (semantics). In this case, you have the coordinate with (0,1) added, and (0,-1) added, to create two coordinates, and all that was happening here was you were subtracting the larger coordinate from the lesser coordinate, instead of the other way around.