Implicit conversion from float -10.603448275862064 to int loses precision