Implicit conversion from float -26.966292134831463 to int loses precision