Implicit conversion from float -143.73015873015873 to int loses precision