I don’t know if they work for predicting the actual election but as a measure of change from before a debate to after a debate I think it’s probably a good way to evaluate debate performance.
And I think prediction markets have wider use now than 2016. More diverse opinions being represented now than before.