Implicit conversion from float -27.94354838709677 to int loses precision