Optimal model:
  x = -32768 :: Int16
  m = -32768 :: Int16