Implicit conversion from float -439.4117647058823 to int loses precision