Implicit conversion from float -77.61904761904762 to int loses precision