Implicit conversion from float -9.557522123893804 to int loses precision