Implicit conversion from float -165.86353944562902 to int loses precision