Implicit conversion from float -15.540720961281707 to int loses precision