Implicit conversion from float -50.03407155025553 to int loses precision