Implicit conversion from float -22.93888166449935 to int loses precision