Hello, overseb
Reversed matrix is used for optimization.
The basic point of this technique lies in swapping Far and Near values in projection matrix.
When depth is calculated for a standard matrix, an addition of numbers of different orders may occur (for example, 50000 and 0.5). This operation is quite heavy due to specifics floating-point numbers representation. On the other hand, when reversed projection matrix is used, this behavior occurs only in the area, that is close to the camera. Thus, it is more effective.
Hope this helped!
Cheers!