The key argument, as I understand it, of Francis Watson's Paul and the Hermeneutics of Faith is that Paul did not have his own free-floating, independent theological system, which he then went to the Hebrew Bible to validate; rather his theology was itself generated by Scripture, especially Torah. Pauline writings are in fact an interpretative venture (interpretation of the Torah). The direction is OT to Paul, not Paul to OT.
Specifically, Watson argues that Paul was an interpreter of the Torah just as his Jewish contemporaries were, but they read the same Scripture differently. Hence to read Paul is to pursue his hermeneutics - the Pauline theological task is fundamentally a hermeneutical one - and Paul specifically operated with a hermeneutics of faith, particularly rooted in Hab 2:4. The task of understanding Paul therefore involves a threefold field of intertextuality, as Watson calls it - involving Paul, his contemporaries, and the OT.
One of the things I've appreciated as I read this book is the way Watson takes Luther seriously - and when he disagrees, he cites original Luther writings, not secondary sources or well-worn accusations about how Luther read a sixteenth century Catholic meritocracy into first-century Judaism and Paul. For example, he closes one chapter by saying:
[I]t is not part of my intention (conscious, subconscious, or whatever) to "rehabilitate Luther." Many of the reasons for the current anti-Lutheran climate in Pauline studies are good ones, and the retractations I would now make of things I said on this matter in an earlier book (Paul, Judaism and the Gentiles, 1986) are few and qualified. Yet there is a potential danger that certain exegetical possibilities will be ruled out in advance merely because of their apparent proximity to Luther. Such an attitude is more suited to some heresy-hunting orthodoxy than to a critical scholarly discipline. (29)
The only concern I register so far is that Watson at times attributes to the law what ought to be attributed to sin. That is, I believe at times he ought to attribute to human failure the impotence of the law to secure human righteousness, and not some defect in the law itself (e.g. 57).