Implicit conversion from float -100.06622516556291 to int loses precision