Implicit conversion from float -246.73469387755102 to int loses precision