Implicit conversion from float -8.503937007874015 to int loses precision