Thank you for raising this. It is confusing behaviour and something that we will be addressing shortly.
Until we ship 10.8, the ratings are still based upon the type of the issue (vulnerability, bug, code smell). Reliability, for instance, is calculated based on the count of ‘bugs’ you have. This can mean you get a situation like the one you are seeing. You can have issues that are ‘code smells’ and so aren’t counted in the reliability rating, but have a reliability quality associated with them and so appear in the count of reliability issues. This is confusing and we will be addressing this in the upcoming 10.8 release which is targeted for December.
If you list the 12 open issues and they are listed as ‘bugs’ and not ‘code smells’ then there is something else going on. Please let us know if that is the case.