Implicit conversion from float -6.048387096774178 to int loses precision