Implicit conversion from float -864.0845070422536 to int loses precision