Implicit conversion from float -45.08823529411765 to int loses precision