In reading his analysis of the election here, he repeatedly mentions that having only two candidates makes things much more predictable.

In any case, we don't know if he got it "wrong" unless we know what % chance he gave this outcome. (He tries to account for the likelihood that polls are misleadingly biased, and that would correlate across multiple seats.)