Implicit conversion from float -116.2937062937063 to int loses precision