Implicit conversion from float -46.02739726027397 to int loses precision