Implicit conversion from float -10.095328884652048 to int loses precision