Implicit conversion from float -45.50420168067228 to int loses precision