Last season, the Mets' bullpen was the weakest link on a team that was supposed to be playoff bound from the get-go. As a unit, Mets relievers were 22nd in baseball with a 4.40 FIP, which was worse than also-rans like the Marlins, Nationals, Royals and Mariners, to name a few. Some of the parts worked reasonably well. Billy Wagner was effective before elbow surgery ended his season prematurely. Brian Stokes was a standout late in the season when everyone else was crumbling around him. For the most part, though, the bullpen was dreadful, as Aaron Heilman, Luis Ayala, Duaner Sanchez and others conspired help pull the Mets down from within, ultimately falling just one game short of the Wild Card.

This year, while the Mets' offense has struggled and their roster has suffered injuries galore, the bullpen has been a bright spot, currently sporting the second-best FIP in baseball at 3.71. Where has the improvement come from? Francisco Rodriguez has certainly helped, walking a few too many batters but otherwise doing everything you expect a relief ace to do. Bobby Parnell has been tremendous, and will see some more high-leverage innings now that he has been promoted to set up Rodriguez. But there's one other new face that has helped the Mets almost as much as those two.

"Mr. Superstar" is Pedro Feliciano. "Bum He Replaced" is also Pedro Feliciano, but his 2008 version. Value-wise, Pedro v2009 has been worth twice as much as v2008 (.6 WAR versus .3 WAR) in roughly two-fifths of the innings. So what happened? A few years ago our analysis would have been limited to pure speculation, distilling what we could from the countless random and often contradictory things a pitcher's manager or pitching coach might have said. Now, with PITCHf/x, we evolved from idle speculation to -- at the very least -- some form of reasonable evidence-based speculation.

In addition to PITCHf/x data, we also have pitch type breakdowns courtesy of FanGraphs. More data is always better as long as the quality of the data renders it meaningful. The first place we can look for clues is in Feliciano's pitch selection to see if he's throwing more of something and/or less of something else. There is some disagreement between the two sources of data: PITCHf/x has Feliciano throwing 18.2% curveballs in 2008, while FanGraphs (really, Baseball Info Solutions, or BIS) has him rarely throwing that pitch (.2%). The primary disparity stems from the way the two systems classify pitches, though it seems pretty clear that an adjustment was made to the PITCHf/x system since last year that is now correctly identifying Feliciano's breaking pitch as a slider, not a curveball. See:

PITCHf/x is still a fairly new system, so performance and accuracy improvements can be expected over time. It's obvious that Feliciano isn't using his slider with any greater frequency in 2009, but it's equally obvious that his other pitches -- his fastball and changeup -- are being apportioned differently this year. Here is PITCHf/x:

And here is BIS:

Regardless of which system you believe, Feliciano is throwing far more changeups this year, a trend that has surely made him less predictable to hitters who might have otherwise been able to sit fastball in fastball counts and slider in slider counts. The changeup splits the difference and gives Feliciano another option in both hitter's and pitcher's counts.

That Feliciano is throwing his changeup more often is not mere happenstance; the horizontal movement of the pitch has doubled over 2008 (5.4 to 10.3), so the pitch clearly has more bite now. What's the sense of improving one of your pitches if you're not going to use it more? The corollary to the movement increase is that Feliciano's O-Swing% (percentage of pitches swung at outside the strike zone) has increased 50% -- from 20.3% to 30.9% -- over last season. Again, no rocket surgery required here. More break on the pitch means more swings and misses.

Another thing that gives Feliciano's changeup a little more juice is that he's throwing his other pitches a bit faster than he did in 2008. His fastball has moved up a tick while his slider has increased anywhere from two (PITCHf/x) to five (BIS) ticks. And lastly, Feliciano is doing a better job of getting ahead of hitters, throwing first pitch strikes 63.1% of the time compared to 53.6% in 2008.

Could Feliciano's second (or third?) life be all a credit to his new-and-improved changeup? There plenty of compelling evidence to suggest so. Whatever the reason, he's a whole lot better than that bum he replaced.