I think the idea is that nobody else will have picked them, so even though it is a fairly small % chance they do well (I made it about 10% chance of clean sheet), the top-heavy payout structure makes it worthwhile, because on those occasions that they do perform well (and other popular defences disappoint), you have a real chance of a very high finish.
As a simplified and exaggerated example, say that there are 100 lineups entered into a tournament. 99 of them have chosen a Liverpool defence which is expected to perform better than West Ham say 90% of the time, and the other lineup has a West Ham defence. Both teams have the same attacking players.
90% of the time the West Ham team finishes dead last, but 10% of the time it wins the whole thing, which is obviously a massively +$EV proposition in a tournament with top-heavy payouts, despite West Ham players being bad picks in terms of pure points EV.
Having said all that, the week I binked my best team had a Man City defence