Implicit conversion from float -53.532684283727406 to int loses precision