Implicit conversion from float -52.65734265734264 to int loses precision