Implicit conversion from float -82.94117647058823 to int loses precision