WPA is nobody's idea of a perfect stat, and is more about situations than about "how good" a given player is. I do like a couple of things about it, one of which is that it tracks well with the fan psychology. The way WPA works, a 9th inning meltdown is more costly than one occurring in the 8th, since confidence in a lead increases as the end of the game approaches.
That's what makes Pagan's most recent loss especially noteworthy. I looked on b-r.com's Stathead tool the next day, and his was only about the 39th worst (negative) game-WPA for a pitcher this season. But 31 of the ones farther up the list occurred in the 9th inning.
Special recognition must go to Josh Fleming for his outing against the Dodgers on May 28. You may remember the game. He started and pitched 6 full innings, at which point the score was tied 10-10. His Rays kept scoring runs, piling up positive WPA totaling .849 by game's end, and he just kept giving runs right back (assisted a bit by defensive mistakes), racking up a total of -.713 WPA. Fleming's number is really hard for a starter to achieve, worst for a starter so far this year. Since the Rays were the home team, and it was tied, they still had an estimated 57% historical chance of winning after he finished the top of the 6th, and they did push across a run for an 11-10 thriller; his two relievers justly earned nice WPA ratings. Weird game.