I was researching this myself earlier today, since my Y12 daughter has just received her ALIS predictions as well.
Her school bases the ALIS predictions purely on GCSE results. Basically ALIS takes a child's results, comes up with an average GCSE "score" and uses that alongside its database of previous students to predict how your child will do at AS/A2 based on how those other students went on to perform.
There is an option to do online testing as well, but my impression from my (admittedly only 10 minutes) internet trawling on the subject was that this is useful only when the student doesn't have a traditional set of GCSEs available, which I can't imagine is the case with the children mentioned in this thread so far.
From what my daughter has observed from her and her friends' predictions, ALIS is OK if you got grades at GCSE that were all the same or very similar. Say, however, you have a science/maths leaning and did really well in those and less well in the arty subjects, the lower grades at GCSE (even if now irrelevant if studying science/maths at A Level) bring down the predicted grades for AS (and the same applies the other way round, of course, for those who are arts-inclined but not as strong on the science side).
My daughter's school uses ALIS as informative rather than set in stone and it is used for initial target setting, which is then obviously assessed by the teachers in line with actual performance during Y12.