Implicit conversion from float -9.92236024844722 to int loses precision