Implicit conversion from float -94.27672955974842 to int loses precision