It's possible to mine data off the web and get a pretty reasonable ballpark value for a domain. Identifying and enumerating the various components that make a domain valuable is a requirement for domaining success -- while most domainers do this by "gut" from similar sales, you don't have to. My valuation models usually start with a domain name and a spreadsheet of various parameters -- keep adding useful info about the domain and tweak numbers as values change. When I find that I am no longer adding new classes of parameters, but only changing the data within the cells, it's time to see if the various information pieces can be distilled into a set of equations. When doing this, I suggest using a operations research / constraint programming perspective. Whether your model will have any generic applicability will depend on your modeling talent. There's nothing wrong with trying to quantify a domain's value. Overall, though, if you get good at valuation modeling, you will probably find that your skills are better suited for more interesting and lucrative markets elsewhere.