Could predicted pagerank be wrong? Is it based solely on number of backlinks, which are not that numerous, or quality?
Don’t know. What I do know is that:
1) Pagerank (as reported by Google to end-users) is pretty meaningless anyway. It’s some general indication of quality, i.e. a PR 6 is probably a better site than a PR3, but the PR reported by Google is not the same PR used internally by Google’s algorithms. You might find a PR6 site that languishes on page 7 of the SERPs, and a PR0 or PR1 site that is ranking #1 for a competitive term.
2) With backlinks, quality is everything (definitely for Google, less so for Y and M). 10 links from .edus or CNN could be better than 1000 or 10,000 from new or spammy sites. I don’t see how a PR predictor could take into account quality, therefore I think PR predictors must be of limited use.
So if your site has high quality backlinks, you may do better than a PR predictor might expect.
But the only real proof is in the results; if you’re already ranking for some decent keywords then you’re in a good position and that is much more important than what Google reports your PR as.