Implicit conversion from float -9.532710280373834 to int loses precision