Implicit conversion from float -69.62566844919786 to int loses precision