Implicit conversion from float -100.24390243902437 to int loses precision