Implicit conversion from float -69.6590909090909 to int loses precision