I basically use an the raw trials and rawmarks.info database to estimate the aligned marks. Then plug this into an ATAR calculator (or if I think externals are different to internals, average them first). I may make a few intuitive decisions and make a few assumptions such as trials were an equal level to the HSC exam and that the performance of the cohort is reflected accurately and they will perform about the same in the externals.
I also double check as well with another method which I think i did for this person. Basically I sorta make a rough estimate on their overall ranks, if I had to give one to the student. I estimated she was around 10th in her cohort. Then there are about 600 schools in the state so I find out how many people is either 10th or above in their schools which is just 10 * 600 = 6000. Then due to lower ranked schools there may be about 1000 people that although ranked at 10 or lower actually may not be as good. This means about 5000 people are ahead of them (though this may be a loose number). Then I eliminated another 1000 just due to maybe kids who aren't as good or who may fail in their exams and assuming this student either performs as good or better in HSC externals (though this may be a bad assumption as there may be a similar amount excelling or doing better). This means there are about 4000 kids who are better than this student, and as each ATAR point is ~ 1000 (actually more like 900) candidates this is about 4 points behind the maximum = 100-4 = 96 roughly. This was similar to my calculator estimate.
The second method may be more informal and may just be my method but it seemed to work well for this case.
