Implicit conversion from float -16.38297872340425 to int loses precision