Implicit conversion from float -133.05194805194805 to int loses precision