Implicit conversion from float -78.08035714285714 to int loses precision