Implicit conversion from float -9.874371859296488 to int loses precision