Implicit conversion from float -52.37288135593221 to int loses precision