Win Expectancy is based off averages. That has a downside in not allowing us to tailor the WE of a situation for the particular batter/pitcher match up, but it also has some advantages beyond that stated ideal being more complex to calculate than the human genome. Because the WEs are based on averages, we would expect an exactly league average player to end up with a WPA of 0 over a large enough sample. And not just single players, but entire units of players.
The Mariners' bullpen has been worth -2.1 runs to average according to tRA. With an average leverage of 1.14, that becomes -2.4 leveraged runs below average. Dividing that figure by the average Mariner run environment, which is actually below 10 runs per win as abender20 pointed out, we would expect the Mariners' bullpen to have a net WPA around -0.3. Not coincidentally, their WPA/LI is at -0.3 as well. WPA/LI is a different route to the same destination.
The bullpen's WPA is at -2.13 instead. That's a 1.8 win difference that could largely be chalked up to serving up bad results at particularly bad times. Call it not being clutch, bad luck, whatever, but it's there.
This isn't addressing the performance of any individual relievers, just saying that it could be stated that the bullpen has been unlucky to almost two wins this year. Of course, this could be extended to the rotation just as easily where running the same calculations yields a group of pitchers that have been clutch/lucky/whatever by about half a win.
I don't think either result is surprising. The bullpen hasn't been as good as we wanted, but given the amount of one-run losses we've endured they are bound to be over blamed by wins, losses, saves and WPA. And the rotation has been good at keeping runs off the board, but tRA isn't as keen on some of the fundamental numbers and sees a rotation worse than its' ERA. With the two combined, it is yet another indicator that the Seattle Mariners have suffered from some poor luck this season.