Implicit conversion from float -15.02035278154682 to int loses precision