Implicit conversion from float -125.9090909090909 to int loses precision