Implicit conversion from float -60.925925925925924 to int loses precision