A good R2 tells you that the assay has good linearity and presumably good precision, but it doesn’t tell anything about accuracy.
All the values could be associated with a systematic bias (bad accuracy) and you would still be getting a good correlation.
If you are lucky and have ‚the correct‘ SHBG level (not knowing what the level is) for the test it could indeed reflect your true freeT levels, but keep in mind that SHBG levels typically change on TRT. If not a bias is introduced as the assay apparently measures a constant fraction of total T independent of SHBG levels. This is also reflected in the very different concentration levels the assay reports.
Fritz et al approached the assay in a very systematic way. Very interesting read. I hope it’s freely accessible.
https://academic.oup.com/clinchem/article/54/3/512/5628433