Implicit conversion from float -465.38461538461536 to int loses precision