Implicit conversion from float -2.4390243902439153 to int loses precision