Implicit conversion from float -14.119850187265925 to int loses precision