Implicit conversion from float -9.476439790575924 to int loses precision