Implicit conversion from float -14.268292682926827 to int loses precision