Implicit conversion from float -11.794871794871796 to int loses precision